Author’s Accepted Manuscript

A Model for Hybrid Bearing Life with Surface and

Subsurface Survival

Antonio Gabelli, Guillermo E. Morales-Espejel

www.elsevier.com/locate/wear

PII: S0043-1648(18)31289-4
DOI: https://doi.org/10.1016/j.wear.2019.01.050
Reference: WEA102719
To appear in: Wear
Received date: 15 October 2018
Revised date: 8 January 2019
Accepted date: 8 January 2019
Cite this article as: Antonio Gabelli and Guillermo E. Morales-Espejel, A Model
for Hybrid Bearing Life with Surface and Subsurface Survival, Wear,
https://doi.org/10.1016/j.wear.2019.01.050
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
 A Model for Hybrid Bearing Life with
Surface and Subsurface Survival

Antonio Gabelli1, Guillermo E. Morales-Espejel1,2,

1
SKF Research and Technology Development, Nieuwegein, The Netherlands
2
Université de Lyon, INSA-Lyon, CNRS, LaMCoS UMR5259, F69621, France

Abstract:
Hybrid bearings have rings made of steel and rolling elements made of bearings
grade silicon nitride ceramic. Due to the higher modulus of elasticity of ceramic
compared to steel, the ellipse of the Hertzian contact is slightly smaller than that in a
conventional bearing. This results in about a 12% increase in the contact pressure
between the ceramic ball and the raceway. Following the standard Rolling Contact
Fatigue theory, a higher contact pressure leads to a reduction of the dynamic load-
carrying capacity and service life of the bearing. However, long experience in the
use of hybrid bearings shows that, in most applications, hybrid bearings clearly
outperform traditional bearings made entirely out of steel.
This paper addresses this issue by applying a novel approach to Rolling Contact
Fatigue. Central to the new method is the survival probability of the raceway surface
which is explicitly formulated into the basic life equations of the rolling contact. This
allows tailoring the contribution of the stress system close to the rolling surface to
better represent the ceramic-steel interaction, which has been proven to be
substantially more favourable in the case of a hybrid ceramic-steel contact. The
comparison between experimentally obtained hybrid bearings fatigue lives and lives
predicted using the new calculation model indicates good agreement.
It is found that an increase of the fatigue resistance of the raceway surface of hybrid
bearings can, in most cases, compensate for the additional stress present in the
subsurface region of the rolling contact. The new methodology, by providing a direct
account of the survival probability of the ceramic-steel interface gives a better
representation of the expected performance of hybrid bearings, opening new
opportunities for the reliable use of this type of bearing in modern machinery.

Nomenclature
2
Surface integration area [m ]
̅ Damage integral constant for the volume [-]
̅ Damage integral constant for the surface [-]

Page 1 of 31
 Exponent in the bearing life equation, and constant [-]
Dynamic load rating (subsurface) [N]
External diameter of the bearing [mm]
Diameter of the rolling element [mm]
Bore diameter of the bearing [mm]
Mean diameter of the bearing [mm]
Exponent in the bearing life equation (standardised Weibull slope) [-]
Exponent in the bearing life equation, [-]
Damage integral for the surface, Index of surface stress [-]

Bearing life rating with 90% reliability [Mrev]

Lower Bound of the 90% two-sided confidence interval on L10 [Mrev]

Upper Bound of the 90% two-sided confidence interval on L10 [Mrev]

Median L10 Estimate [Mrev]

Bearing life rating with 50% reliability [Mrev]

Number of fatigue stress cycles [-]
Bearing rotational speed [rpm]
Equivalent load in the bearing [N]
Fatigue load limit of the bearing [N]
Exponent in the bearing life equation [-]
PDF Probability Density Function [-]
Number of stress cycles per revolution [-]
Precision of an L10 estimation (ratio L10,95% / L10,5%) [-]
Reliability [-]
SiL Significance Level: Quantitative assessment for comparison between parameter
estimations from test data sets [-]
3
Integration volume [m ]
Coordinate (depth from surface) [m]
Number of rolling elements [-]
Greek Symbols
Raceway contact angle [-]
β Weibull shape factor [-]
Lubrication quality factor [-]
Contamination factor [-]
Environmental factor [-]
Viscosity ratio in the bearing [-]
Fatigue generating stress [Pa]
Fatigue limit [Pa]

Page 2 of 31
Subscripts

Refers to hybrid bearing [-]

Refers to steel-steel bearing [-]

Refers to volume [-]

S, A Refers to surface [-]

1 INTRODUCTION
Hybrid ceramic bearings are typically used for demanding operating conditions and
challenging applications. This is because of the better fatigue performance and
reliability that can be achieved using this type of bearing.
For many years hybrid ceramic bearings have been used in high-speed machine
tool spindles to exploit the good speed capabilities and high stiffness of this type of
bearing. In the last few years, the use of hybrid bearings has also grown in many
other industrial applications, with a market that is presently outgrowing the traditional
machine tool sector. The expectation is that this trend will continue and that many
other modern industrial applications will discover and take advantage of the unique
capabilities offered by hybrid ceramic bearings.
Despite the great progress in fatigue performance of ceramics, no clear
performance advantage is recognized to hybrid bearings from the traditional Rolling
Contact Fatigue methodology used in the bearing industry. A realistic forecast of the
expected performance of hybrid bearings is needed for the further growth of the use
of this type of bearing throughout the industry.

1.1 THE DEVELOPMENT OF SILICON NITRIDE HYBRID BEARINGS

1.1.1 Early developments
The use of ceramics, as bearing material, was first proposed in the 1960s for
extreme temperature bearings for aero and space applications Zaretsky and
Anderson (1961) [1], Parker et al. (1965) [2].
Using component Rolling Contact Fatigue testing, Baumgartner (1973) [3], Parker et
al. (1975) [4], Delal and Chiu (1975) [5], established that fully dense hot-pressed
silicon nitride (Si3N4) have the best Rolling Contact Fatigue characteristics when
compared to other ceramic materials. Significant variations in fatigue performance
were also experienced from seemingly identical production batches of silicon nitride
balls, Parker et al. (1975) [4].
In the 1980s Lorösch et al. [6], performed fatigue life testing of angular contact ball
bearings. Lorösch, using the best available quality of silicon nitride balls, found that

Page 3 of 31
hybrid bearing Rolling Contact Fatigue resistance is comparable to the one of all-
steel bearings tested under the same load. This despite the 12% increase in contact
pressure experienced in the hybrid bearings. However, an additional test of a
second batch of silicon nitride balls produced poor fatigue life indicating the criticality
of the quality of the ceramic balls in achieving good fatigue life performance.

Miner et al. (1981) [7], performed functional tests of gas-turbines and found that
hybrid bearings perform significantly better than the all-steel bearings under
lubricant starvation developed during oil shut-off tests. An extensive review of silicon
nitride Rolling Contact Fatigue was performed by Bhushan and Sibley (1981) [8].
They found that in general the fatigue performance of hybrid bearings is like the
nominal rated life of standard steel bearings used at that time.
Morrison et al. (1984) [9] performed four endurance life tests of angular contact
bearings with hot-pressed silicon nitride rolling elements. They found that hybrid
bearing fatigue life has a higher load-life exponent and failure rate comparable to
conventional all-steel bearings. They also observed that the fatigue life was
dominated by spalling fatigue of the ceramic balls giving rise to a lower than
expected fatigue life for hybrid bearings.
The calculation of fatigue life of hybrid bearings for gas turbine engines is discussed
by Zaretsky (1989) [10]. The paper reviews the experimental data of Baumgartner
(1973) [3] and Parker et la. (1975) [4] and concludes that bearings with silicon nitride
rolling elements should be rated at best at 1/10 of their dynamic load rating, (a
thousand-fold reduction of their nominal rated life). Chiu and Tallian (1989) [10], in
the discussion of the Zaretsky’s paper, argued that the experimental data used by
the author were obsolete and not representative of the current performance of
silicon nitride hybrid bearings. Later, Zaretsky et al. (2005) [11] revisited the subject
of the effect of silicon nitride in the bearing life calculations. In this new paper the
authors, applying Weibull [12] and Lundberg-Palmgren [13] [14], types of
exponential life equations, evaluates the effect on the nominal calculated life, of the
higher modulus of elasticity of silicon nitrite. They concluded that the life of hybrid
bearings should be penalized by a factor about two in case of ball and a factor about
three in case of roller bearings.
The inconsistent fatigue test results of hot-pressed silicon nitride balls observed by
Parker et la. (1975) [4], Delal and Chiu (1975) [5] and Lorösch et al. (1980) [6], and
the conservative approach for the estimation of the expected fatigue life of Zaretsky
(1989) [10] are to be viewed considering the limitation in material purity and
manufacturing technology of silicon nitride balls used in the seventies and eighties.

1.1.2 Later developments

Silicon nitride ball quality and performance is discussed by Ebert (1990) [15] using
endurance test results of hybrid bearings with “good” and “poor” quality silicon
nitride rolling elements. Ebert found that bearings with “good” quality silicon nitride
can provide performance comparable to conventional steel bearings. However, the
Rolling Contact Fatigue of “poor” quality (but seemingly identical) silicon nitride balls,
were found to be two to three orders of magnitude less reliable. The very large
spread of RCF of silicon nitride balls experienced in the eighties is due to two main

Page 4 of 31
issues that hampered bearing grade silicon nitride development in the seventies and
eighties:
a) the lower fracture toughness of ceramics that makes silicon nitride rolling
elements more sensitive to tensile stresses developed near cracks and surface
defects.
b) the lack of non-destructive inspection methods to assure uniformity of the
material quality and consistent RCF performance of balls and rollers that are
produced.
The issue of silicon nitride RCF performance and quality assurance was addressed
in the late 1980 and 1990 by Cundill (1990) [16]. Cundill developed proprietary non-
destructive quality rating technology (Non-Destructive Evaluation, NDE) for bearing
grade silicon nitride balls that was introduced in industrial hybrid bearing production
in the early 1990s, Cundill (1993) [17].
The introduction of Non-Destructive Evaluation NDE methods combined with the
continuous progress in silicon nitride material purity and sintering technology finally
allowed engineers to develop reliable and consistent RCF performance of silicon
nitride balls, see Galbato et al. (1992) [18]. This allowed, in the 1990s, to have a
substantial growth in the use of hybrid bearings in high speed machine tool spindles
with significant benefits for the speed capability and precision of these mechanical
components, Cundill (1993) [17].
Cundill (1990) [16] found that the implementation of NDE had a strong impact on the
RCF of the ceramic rolling elements. He shows [16] that silicon nitride ball fatigue
strength could outperform the one of steel ball used in aero-engine bearings. Note
that the tests were performed under the same load, thus the ceramic balls were
subjected to a higher contact pressure compared to the steel variant.
Using the same RCF testing methodology, Cundill (1990) [16], found that a
presence of porosity and surface defects in the material could strongly affect the
RCF of the ceramic rolling elements. These findings further emphasized the need of
reliable non-destructive quality assessment of silicon nitride material to be able to
fully exploit the performance advantages given by silicon nitride hybrid bearings.
In the 1990s the systematic use of NDE made it possible to have consistent high-
quality silicon nitride rolling elements enabling the use of hybrid bearings in several
environmentally challenging applications. Chiu et al. (1996) [19] reported results of
hybrid bearings tested at a contact pressure of 3.5 GPa. The tests provided RCF life
well in excess than the rated life of all-steel baseline bearings. Dezzani et al. (1996)
[20] performed a direct comparison of angular contact hybrid and all-steel bearings
tested under the same loading and application-like lubrication conditions. The test
results clearly indicated a significant fatigue life advantage for the hybrid variant
despite the higher contact pressure developed in hybrid bearings during the test.
A remarkable result of this test was that not even a single ball failure was observed
on a population of 180 silicon nitride balls that were endurance-tested for 1200
Million revolutions, (corresponding to 2000 hour running time). The Dezzani's tests
were performed with angular contact ball bearings, thrust loaded to a contact

Page 5 of 31
pressure of 2.6 GPa. Comparison of Dezzani’s test results (1996) [20] with those of
Morrison et al. (1984) [9] shows the very significant progress in fatigue performance
of silicon nitride balls that occurred between 1984 and 1996. Indeed, in the 60
angular contact ball bearings endurance tested by Morrison et al. [9] practically all
failures were caused by spalling of the silicon nitride balls. More specifically in the
2.4 GPa contact pressure test, 7 out of 63 silicon nitride balls failed. All ball failures
occurred within 76 Mrev giving a 90% survival probability of just 15 Mrev.
Comparing this ball life with Dezzani’s ball survival results it is found that the RCF of
silicon nitride balls increased more than hundred times, in the period between 1984
and 1996.
More recently, Forster at al. (2017) [21], reported endurance test results
accumulated after Rosado et. al. (2009) [42] publications. Here after, these
endurance test results will be referred at as: Rosado-Forster [42] [21]. The Rosado-
Forster’s test results are related to endurance testing of angular contact ball
bearings, hybrid and all-steel variants, performed under the same bearing load thus
inducing a contact pressure of 3.5 GPa and 3.1 GPa respectively in the two variants.
Also, in this case no single silicon nitride ball failure was reported, even though the
tests suspension time was extended to 5000 hours.
Miner et al. (1996) [22] and Shoda et al. (1997) [23] carried out functional tests of
hybrid bearings for aero engine applications. Their results confirmed the ability of
hybrid bearings to withstand oil shut-off tests extremely well. The ability of hybrid
bearings to be unaffected by extreme oil starvation conditions was found to be a
unique feature of hybrid bearings. Jeng and Huang [24] (2000) measured the
temperature rise of hybrid bearings operating at high speeds with oil-mist lubrication.
They found that hybrid ball bearing always had a lower temperature rise compared
to the all-steel variant. The performance advantage offered by hybrid bearings
operating under particles contamination is discussed by Wan et al. [25]. The paper
presents test results of the measure of wear rate of hybrid and all-steel bearings.
Wan found that hybrid bearings have a high wear resistance compared to the
different all-steel variants that were tested.
In the last twenty years we have seen the increased use of hybrid bearings in
challenging applications. This occurs because conventional all-steel bearings
solutions are found unable to meet the reliability requirements of certain difficult
applications. On the other hand, the use of the hybrid bearing solutions is in general
able to resolve the issue. One of such applications is discussed by Morales-Espejel
et al. (2004 [26] and Wallin et al. (2011) [27] about the implementation of hybrid
bearings in large industrial refrigerant compressors. In this application the bearing
operates under very tough lubrication and contamination conditions due to the
presence of refrigerant fluid in the bearing. Hybrid bearings were found able to
provide the long service life required from this demanding application.

1.1.3 Recent developments - tribology of hybrid contacts

To have a better understanding of hybrid bearing performance, researchers have
recently concentrated their efforts on the in-depth analysis of the tribology of
ceramic-steel contacts. Brizmer et al. (2015) [28] investigated the mechanism of

Page 6 of 31
micropitting initiation in hybrid rolling contacts. The authors, using the Morales-
Espejel et al. (2011) [29] surface fatigue calculation model, found that the high
micropitting resistance of the ceramic-steel contact is, for a large extent, the result of
the lower friction coefficient developed in the ceramic-steel interface under boundary
lubrication conditions. In the investigation, the boundary friction coefficient as
measured by Hager et al. (2011) [30] was adopted. The predicted micropitting
resistance for both steel-on-steel and hybrid contacts was found to match well the
extent of micropitting damage observed in dedicated rolling bearings tests. This
work provided a good understanding of the tribology and surface fatigue developed
in hybrid contacts and gives a clear explanation of the performance observed in
hybrid bearings during aero engine oil shut-off tests as reported in [22] and [23].
Vieillard et al. (2016) [31], following a similar research strategy, performed an
extensive investigation of the mechanism of crack initiation near surface defects and
dents present in hybrid bearings that are exposed to solid particle contamination.
Using advanced numerical simulations [33], and dedicated experiments, [32],
Vieillard found the tribological explanation of the improved resistance of hybrid
contacts to the initiation of the fatigue damage around surface defects and dents.
Vieillard validated the theory by carrying out endurance testing of ball bearings with
pre-dented raceways. Vieillard’s endurance test results indicated that hybrid
bearings surface-initiated spalling is significantly delayed compared to what
observed in equivalent tests of bearings made entirely in steel.

1.2 CONCLUSIONS AND SCOPE OF THE PRESENT WORK

Figure 1 – Endurance life of hybrid and steel 7208 bearings, tested at a maximum Hertzian pressure of
3.5 GPa and 3.1 GPa respectively, and with good lubrication.

Page 7 of 31
Figure 2 – Endurance life of hybrid and steel 7208 bearings, tested at maximum Hertzian pressure of
2.6 GPa and 2.3 GPa and under challenging lubrication conditions.

From the review of hybrid bearing development, it is found that the fatigue strength
of bearing grade silicon nitride has made tremendous strides during the last twenty
years. RCF performance of silicon nitride balls have risen of two orders of
magnitude since the early 1980s. Fatigue life testing of hybrid bearings also shows
unequivocally that the ceramic rolling element is the most reliable component in a
bearing system. Nevertheless, the impact of 12% higher contact pressure present in
hybrid bearings is a characteristic that can be observed in current bearing life testing.
This particularly under conditions in which subsurface contact fatigue is the
predominant damage mechanism controlling the fatigue life of the bearing. Figure 1
shows the Weibull probability plot of the Rosado-Forster’s endurance tests of hybrid
and all-steel bearings tested under the same loading conditions. The plot shows that
under high load and favourable lubrication conditions, subsurface fatigue dominates
the bearing fatigue performance. It follows that hybrid bearings that are subjected to
a high contact pressure (3.5 GPa) generate a higher failure probability, for a given
running time, compared to the all-steel bearings tested under the same load, but
under a lower contact pressure (3.1 GPa).
Figure 2 shows the Weibull probability plot of the endurance tests performed by
Chiu et al. [19] comparing all-steel and hybrid executions under the same test
conditions. The tests were performed at a lower load giving rise to a maximum
Hertzian contact pressure of 2.6 GPa for the hybrid and 2.3 GPa for the steel variant.
The tests were performed under challenging lubrication conditions. High
temperature, thin film and raceways run-in with oil contaminated with hard particles
to simulate the actual lubrication condition of the application.
Under the given conditions, bearing life is dominated by surface-initiated fatigue.
Hybrid bearings of Figure 2 show a clear performance advantage with a substantial

Page 8 of 31
increase of RCF life, even if they were subjected to a higher contact pressure. The
statistical confidence of this test result is high.
Chiu’s test results cannot be explained with the traditional bearing life models.
Traditional methods to predict Rolling Contact Fatigue for conventional all-steel
bearings are somehow inadequate to represent the actual performance of hybrid
bearings. Standard RCF calculations are ultimately based on subsurface fatigue and
would always penalize the life of hybrid bearings due to the 12% increase in contact
pressure of the rolling contact. However, this would not provide a fair description of
the actual performance of hybrid bearings when they operate under environmental
conditions found in many bearing applications. Therefore, what is needed is a
fatigue life model that can discloses, and gives full representation, of the superior
tribo-properties of the ceramic-steel interface of hybrid bearings, thereby to the
unique fatigue life performance observed in many hybrids bearing applications.
Forster et al. (2017) [21], discussing their investigation on hybrid bearing fatigue life
recognize the shortcomings of traditional RCF approach. They recommended the
development of novel RCF approach able to include, explicitly, the fatigue properties
of the raceway surface, as proposed by Morales-Espejel et al. (2015) [34]. The
scope of the present paper is to present such new development aimed to improving
the calculation of the Rolling Contact Fatigue performance of hybrid bearings.

2 GENERALIZED ROLLING CONTACT FATIGUE

The detailed description of the Generalized Bearing Life Model (GBLM) is given in
[34]. Herewith only a short outline of the method is presented with emphasis to the
aspects of the model that are related to hybrid bearings.

2.1 PROBABILISTIC FATIGUE DAMAGE

A bearing can be viewed as a system of components, i.e. rolling elements, raceways,
cage, that are connected in series. In a series, or chain configuration all components
are critical to the reliability of the system and its survival is dictated by the product law
of reliability. The survival probability of the complete bearing is therefore provided by
the product of the survival probability of each single component part of the bearing
system.
This general approach is also used for the calculation of the survival probability of
the material stress volume of the rolling contact, [12], [13], [14] and [35]. A short
summary of this methodology is given below.
The probability of survival of a rolling contact can be related to a structure composed
by elements connected in series configuration. The elements are subjected to
different stress states and therefore have different probability of survival:
. In such a system all components are critical if one element of the
structure fails the entire system fails. As mentioned, in a series configuration the
survival probability of the whole structure is given by the product of the survival
probability of each of the elements of the structure,

Page 9 of 31
Similarly, for a rolling contact the survival probability , to endure stress
cycles, is given by the product of the survival probability of each of the
elements constituting the stressed volume of the rolling contact.

(1)

By rewriting equation (1) as a reciprocal function and applying the logarithm rules at
both sides of equation, the following equivalent form is obtained,

[ ] [ ] [ ] [ ] (2)

Let us consider the volume element sufficiently large to contain material defects.
We can introduce a material strength degradation function to account for the
fatigue damage accumulation, at each stress cycle, in the volume element .

The survival probability of equation (2) can then be expressed in term of the fatigue
damage generated by the function after stress cycles in the volume .
Accordingly, for the stressed volume , the survival probability is:

[ ] (3)

The combined effect on the survival of the complete material volume can be derived
from eq. (2) as summation of the elements constituting the stressed volume of
the rolling contact subjected to fatigue cycling. Finally, letting allows to
adopt an equivalent integral form to represent the survival probability of the contact.

[ ] ∑ ⇒ ∫ (4)

In a hybrid rolling contact, the stressed volume can be further subdivided into two
separate sub regions that are the main sources of independent fatigue damage
developed in the rolling contact, see Figure 3.

Using the notation (subscript ) for the subsurface region and (subscript ) for
the surface region of the rolling contact, we can rewrite equation (5) using two
separate terms. The modelling of subsurface fatigue damage can be achieve using
the fatigue damage function while the tribology and degradation processes
taking place at the very surface of the ceramic-steel contact can be achieved using
the damage function that is applied to a very narrow layer, of thickness ̂ , of
raceway surface, see Figure 3. Equation (4) can then be rewritten as:

[ ] ∫ ̂∫ (5)

Page 10 of 31
Following Ioannides et al. (1985) [35] the fatigue damage volume integral of
equation (5) can be expressed as a function of the critical stress amplitude
(fatigue damaging stress) originated within the Hertzian stress field during over-
rolling. The proposed damaging function, according to [35] is:

〈 〉
∫ ̅ ∫ (6)

In equation (6), and are experimental constants that characterize the fatigue
damage generation at each over-rolling cycle, while represents the failure-rate of
the Weibull statistics for subsurface fatigue. is the contact life in number of over-
rolling load cycles, represents the depth of analysis, is the integration volume,
is the fatigue limit stress [36] of the material and ̅ is the bearing subsurface
constant for Rolling Contact Fatigue.
The above methodology is quite general; thus, a similar approach is followed to
represent the surface damage function . If the surface depth constant ̂ , see
Figure 3, is included into the surface RCF constant ̅ , one obtains:

̂∫ ̅ ∫〈 〉 (7)

Where is the area of integration, is the fatigue limit at the surface stressed
layer, ̅ is the surface RCF constant, and is the characteristic slope of the Weibull
statistics of the bearing life for surface originated fatigue failure.
In the surface damage function (7) the surface stress amplitude during over-rolling
is obtained from the stress system present in a layer ̂ of few micrometres depth
below the raceway surface, see Figure 3. Quantification of these stresses must be
obtained from the actual contact surface micro-geometry, i.e. roughness, dents, and
frictional stress condition present at the raceway surface of the hybrid rolling contact.

2.2 GENERALIZED BEARING LIFE FOR HYBRID BEARINGS

Combining equations (6) of the subsurface and (7) of the surface, with equation (5),
it is possible to obtain a generalized life equation with separate terms for the surface
and the subsurface.
Notice that the life of the bearing, in number of revolutions, is related to the
number of over-rolling stress cycles by the ratio , where is the
number of the over-rolling stress cycles taking place at each revolution.
In bearing life ratings, a standardized failure-rate of the Weibull statistics is generally
adopted, then we can set . Furthermore, indicating with the 10% bearing
contact life and setting equal to 0.9 (for 90% survival), we can rewrite (5) as:

Page 11 of 31
 〈 〉
[ ( )] [̅∫ ̅ ∫〈 〉 ] (8)

Equation (8) represents the basis of a Generalized Bearing Life Model that explicitly
separates the fatigue damage of the raceway surface from the subsurface fatigue of
the rolling contact.
The subsurface term of equation (8), (represented by the volume integral), can be
solved using established Rolling Contact Fatigue methods, see reference [35].
However, the surface term, given by the area integral of equation (8), must be
quantified in a radical different manner. Its assessment requires the estimation of
the damage originated by the actual stress conditions of the raceway surface under
a variety of operating conditions that may occur to the bearing.
This task is complex, but it offers the possibility to consistently consider, in the life
estimation of hybrid bearings, the actual tribological phenomena that characterize
the performance and endurance characteristics of the ceramic-steel raceway
contact as extensively described in Brizmer [28] and Viellard [31].

Figure 3 - Schematics of surface and subsurface von Mises stress of a Hertzian bearing contact.

Page 12 of 31
2.3 SURFACE SURVIVAL OF THE CERAMIC-STEEL INTERFACE
Equation (8) can be rewritten in a form that clearly indicates the contribution of the
raceway surface to the L10 life of a rolling bearing contact:

̅ 〈 〉 ̅
∫ ∫〈 〉 (9)
⏟( ) ⏟( )
[ ]

For a given bearing size and excluding constant terms, the surface fatigue damage
of equation (9) is a direct function of the combined effects of the stress conditions
experienced by the raceway surface during the over-rolling of the rolling contact:

∫〈 〉 (10)

The assessment of the surface damage integral (10) can be accomplished by

integrating the surface originated fatigue stress of the raceway resulting from a
variety of operating conditions of the bearing.

Figure 4 - Schematic flowchart of advanced Micro-EHL for surface fatigue stress damage evaluation

In the current formulation, following Morales-Espejel [34], [37], the surface stressing
and damage accumulation can be treated using advanced surface distress

Page 13 of 31
modelling for elastohydrodynamic lubricated rolling-sliding rough contacts, i.e.
micro-EHL.
This numerical approach can handle boundary lubrication conditions and presence
of dents and other surface defects, see [37], [38] and [40]. This methodology
requires the use of 3D area samples of the raceway roughness, including the
presence of possible indentations and surface defects. A schematic flowchart of
advanced Micro-EHL calculation, for surface fatigue stress damage, is shown in
Figure 4.
The use of advanced surface stress analyses based on Micro-EHL can be time
consuming and can be impractical for typical bearing life calculations. Therefore, a
parametric study, was set up, to derive a simplified analytical methodology for the
fast estimation of the surface fatigue damage of hybrid bearings operating under a
variety of different loading and lubrication conditions. For this parametric study
surface topographies of hybrid bearing raceways were collected using 3D laser
scanning. This work involved the measurements of several raceways of deep
groove, angular contact and radial roller bearings. The sampling included bearing
raceways that were run-in under different environmental conditions ranging from
very good lubrication to boundary lubrication and severe contamination.

Figure 5 - Typical Micro-EHL calculation results. Raceway roughness of deep groove ball bearing
6204. Film thickness to composite roughness ratio of the calculation is 2.7. Nominal Hertzian contact
pressure 2.3 GPa: A) 3D Measured microgeometry of the bearing raceway, B) Corresponding elastic

Page 14 of 31
deformation of the roughness within the EHL contact. C) Calculated FFT Micro-EHL surface
pressures, D) a roughness profile and corresponding von Mises stress just below the raceway
surface.

From the collected surface micro-geometries, several numerical fatigue stress-

integration were performed as schematically illustrated in Figure 4. The details of
this type of numerical calculations are explained in [34], [37], [39] and [40]. Herewith
the methodology is illustrated presenting the results of a typical calculation case.
This calculation regards the stress conditions of the raceway of a hybrid deep
groove ball bearing type 6204. Surface roughness samples of the bearing raceway
were measured using a 3D laser scanning microscope. The surface roughness
sample used in the present calculation is shown in Figure 5 A).Under a nominal
Hertzian contact pressure of 2.3 GPa and a film thickness to composite roughness
ratio equal to 2.7, the original surface topography undergoes elastic deformation
while passing across the rolling contact. The calculated elastically deformed surface
topography is presented in Figure 5 B)
During the simulation of the contact over-rolling, the Micro-EHL surface pressures
are calculated using the FFT method. The results of these calculations are shown in
Figure 5 C). Detailed view of the Micro-EHL surface pressures corresponding to the
roughness profile y=0 is given in Figure 5 D). The von Mises stress originated by the
bearing topography just below the raceway is also shown in Figure 5 D).
The above type of calculations of the surface originated pressures and related
fatigue stress integration, was performed parametrically. The parameter study
involved rolling contacts of different raceway topographies that were studied under
different contact pressures and lubrication conditions. In the numerical simulations,
the lubrication conditions of the bearing were modelled performing the transient
Micro-EHL calculation of the rolling contact for different lubricant film thickness. The
effect of lubricant contamination was included by accounting for the presence of
dent topographies and local surface tractions as discussed in [31], [37] and [40].
Although all results discussed here are related specifically to hybrid bearings, similar
surface stress fatigue calculations were performed also for all-steel rolling contacts,
[34], [37]. This to have a base line reference model also for standard all-steel
bearings.
Under mixed lubrication conditions, there are significant benefits to the stresses
developed at the surface of a hybrid bearing raceway, compared to standard all-
steel bearing raceways. Detailed evaluation of the reduction of surface originated
stress and improvement in surface fatigue provided by hybrid bearings under mixed
lubrication conditions are discussed in detail by Brizmer [28] and Viellard [31]. In the
present work, this is further extended with the development of surface stress integral
for hybrid able to account for these benefits into the L10 fatigue life expectation of
the bearings.

Page 15 of 31
2.4 ANALYTICAL MODEL OF THE SURFACE STRESS INTEGRAL
The numerical results of the parametric study of the surface stress integral of hybrid
bearings, discussed in section 2.3, indicated the possibility to represent this quantity
using an analytical formulation. Similarly, as originally described in [34], it was found
that all numerical results could be approximated by an exponential function.
This function depends on two main parameters, the relative load and the
environmental lubrication factor of the bearing application, . The
surface stress integral has the form,

∫〈 〉 [ ] (11)

In equation (11) the parameters are determined numerically as

data sets that are dependent to the bearing type and environmental lubrication
conditions ( of the bearing. Their values are stored in form of data blocks in the
calculation program.
The environmental factor ( used in the derivation of the c parameters of
equation (11) is obtained from the lubrication ) and contamination
factors that are used in bearing engineering practice [39] and [40].
A plot of the surface stress integral as index of surface stress of the
hybrid contact is given in Figure 6. The plot shows that under good lubrication, free
from contamination particles, the expectancy of increased surface stresses, as
indicated by the surface stress index, is indeed negligible. Whereas, under condition
of severe contamination and/or reduced lubrication, surface stress fatigue damage
reaches its maximum value.
Figure 6 shows also that a reduction of environmentally induced stress for the
bearing raceway can be achieved by limiting the applied load of the rolling contact.
This effect is significantly reduced in case of a severe deterioration of the lubrication
of the bearing, i.e. severe environment .

Page 16 of 31
Figure 6 - Index of surface stress of hybrid bearings as function of load and lubrication environment

2.5 GBLM HYBRID BEARING FATIGUE LIFE

Similar to [34], the availability of an analytical form of the surface fatigue stress
integral for hybrid bearings, equation (11), gives the possibility to predict the fatigue
life without the use of time consuming numerical simulations. Inserting equation (11)
in the bearing life model (8) provides a life rating model for a hybrid bearing rolling
contact:

[ ( )] 〈 〉
[̅ ∫ ̅ [ ]] (12)

The Generalized Bearing Life Model (GBLM) of equation (12), integrates the results
of the research work Brizmer [28] and Viellard [31] in its formulation. It allows for the
proper consideration of the damaging mechanisms and specific tribology that
characterize a ceramic-steel rolling contact. By providing a direct account of the
survival probability of the ceramic-steel interface, equation (12) gives a good
representation of the actual performance of hybrid bearings as it will be discussed in
the next section.

3 APPLICATION OF THE MODEL

The Generalized Bearing Life Model of equation (12) can be applied to represent the
endurance test data of Rosado-Forster [21] and Chiu [19] that were discussed

Page 17 of 31
earlier. The tests were performed on a population of 72 angular contact ball
bearings. A total of 40 hybrid bearings were tested leading to 12 failures. For the all-
steel variant, 32 bearing were tested leading to 21 failures. Geometrical details of
the test sample, the loading and stress conditions used in these endurance tests are
shown in Table 1.

Table 1 - Bearing size 7208, hybrid and all-steel variants. geometrical details and test loading.

Parameter Rosado-Forster [21] Chiu [19]

Ball diameter (mm) 12.7 12.7

Number of balls (-) 11 9

Pitch diameter (mm) 60.25 60.25

Nominal contact angle (o) 22 18

Axial load (kN) 22.2 7.56

Hybrid pressure (GPa) 3.5 2.6

Steel pressure (GPa) 3.1 2.3

Environmental factor (-) 0.85 0.035

A main difference between the two tests is related to the loading conditions and
lubrication environment in which the tests were conducted. In case of the Rosado-
Forster’s endurance tests, the load was high, leading to Hertzian contact pressure of
3.5 GPa for the hybrid bearing variant. Furthermore, the lubrication conditions were
good with on line oil filtration and a calculated film thickness to composite roughness
ratio larger than two. The GBLM environmental factor resulting from the given
lubrication conditions is then . This gives rise to a low value of the index of
surface stress. Indeed, under the given loading conditions, subsurface fatigue
controls the performance of the rolling contact. Therefore, the Rosado-Forster’s test
conditions are of advantage to the all-steel bearing variant which operates at a lower
Hertzian stress of 3.1 GPa and will generate a lower amount of fatigue damage per
over-rolling cycle.

In case of Chiu’s endurance testing, the applied load was substantially lower leading
to a maximum Hertzian stress of 2.6 GPa for the hybrid variant. The running
temperature of the tests was higher (150 oC) providing a thinner oil film to the
bearing. Furthermore, the tests were conducted with induced raceway defects to
reproduce the typical contamination conditions encountered in many bearing
applications. This was achieved by running-in the bearings for 15 minutes in oil
containing 2.5 ppm aluminium oxide particles of 20 size.

The resulting GBLM environmental factor that characterizes the above

operating conditions is 0.035 This was obtained with an estimated film thickness to

Page 18 of 31
composite roughness ratio of the test 0.675 corresponding to a lubrication factor,
according to [40], 0.175 and ISO 281 factor for contamination, 0.2.
In case of Chiu’s endurance testing, with the given loading and environmental
conditions, the role of the surface fatigue dominates the survival of the bearing. In
other words, there is a high index of surface stress for the given test conditions.
Thus, surface fatigue will control the fatigue performance of the bearing under test.
The surface stress index specifically developed for hybrid bearings will play here in
favour for the fatigue performance of the hybrid bearing. It will provide a lower rate
of surface fatigue damage for the hybrid raceway, thus compensating for the higher
Hertzian stress (2.6 GPa), present in the ceramic-steel contact.

The running conditions of Chiu’s tests, discussed above, were introduced into an ad
hoc bearing life code with the implementation of the Generalized Bearing Life Model
according to eq. (12) The life performance of the 7208, steel variant, was also
computed using a version of the GBLM suitable for all-steel bearings as described in
Morales-Espejel [34]. The results of the computed ten percentile fatigue life
corresponding to the different tests and bearing variants are presented in the
Weibull plots of Figure 7 and 8. The GBLM predicted L10 endurance lives, in
millions of revolutions, are shown in the plots with a circular symbol positioned at the
10% failure probability of the Weibull plot. To improve the visual clarity of the plot, a
vertical line is added and labelled according to the corresponding bearing variant of
the calculation, Figure 7 and 8.

Figure 7 – Endurance life tested at a maximum Hertzian pressure of 3.5 and 3.1 GPa (hybrid and steel)
and good lubrication. Calculated L10 lives are indicated with a circular symbol.

Page 19 of 31
Figure 8 – Endurance life tested at a maximum Hertzian pressure of 2.6 and 2.3 GPa (hybrid and steel)
and challenging lubrication. Calculated L 10 lives are indicated with a circular symbol.

In the Weibull probability plot of Figure 7 and 8, the test L10 life is shown with its
two-sided 90% confidence interval. The notation L10,5 and L10,95 is used to
indicate the lower (5%) and upper (95%) bound of L10 confidence interval. The
notation L10,50 is used to indicate the median L10 life of the test.

The 90% confidence interval of the Weibull probability plot is related to the
Probability Density Function (PDF) that the true (but unknown) L10 is to be found
within a given life value. It follows that the experiments will confirm a predicted life
value with a statistical significance that is given by the right-side integration of the
PDF starting from the life value that is given.

This implies that, if the predicted (given) L10 life is equal or lower than the lower
bound of the confidence interval of the test (i.e. L10,5), the model result is confirmed
with a high degree of statistical significance. On the other hand, in case the
predicted life L10 approaches the median L10 life of the test, that is L10,50 there is
a reduction of significance.

Considering the above, the plot presented in Figure 7 shows that predicted fatigue
lives for the hybrid and all-steel variants are set at about the lower end, (i.e. L10,5)
of the 90% confidence of the endurance test results. This signifies that the Rosado-
Forster’s endurance tests confirm the GBLM model results with a high degree of
statistical significance.

The model results related to Chiu’s endurance tests are presented in Figure 8. In
this case the predictions have a somewhat reduced statistical significance. This may
be due to the low number of failures obtained for the hybrid variant and to a one
early failure affecting the results of the all-steel test. Nevertheless, the predicted
GBLM lives are in all cases lower than the median L10 life of the test, i.e. L10,50,
thus are acceptable.

Page 20 of 31
The predicted life results show also good consistency with the overall characteristics
of the experimental observations. Indeed, the predicted life results indicated the
ability of the model to represent well the substantial longer Rolling Contact Fatigue
life of the hybrid variant. This despite the higher contact pressure experienced by
this bearing variant, during the life test.

It is to be clarified that the model life predictions and the assessment presented here
are based only on the L10 life evaluation. This is because the L10 life is the
standard life parameter used to assess bearing reliability. Furthermore, the present
life model adopts a standardized Weibull failure rate exponent. Using this standard
Weibull exponent, it would possible to extrapolate the L10 life towards other life
percentiles. This is not new and not part of the present evaluation.

4 MODEL CORRELATION TO ENDURANCE DATA

Further evaluation of the behaviour of the Generalized Bearing Life Model (BLM) for
hybrid bearings can be obtained by checking the model predictions versus
endurance test results produced throughout the years in an industrial research test
facility located in the Netherlands.

4.1 TEST EQUIPMENT AND METHOD

A general description of this endurance test facility is given in [41]. This bearing
testing laboratory is ISO 9000/9001 certified. Thus, full traceability is established,
and procedures are in place for the calibration of the measuring equipment and
execution of the testing activities. Most of the tests were executed on specifically
design test batteries comprising several SKF-R2 endurance life test rigs, see Figure
9.
The SKF-R2 test machines have a simple configuration: single centrally driven shaft
with two support bearings symmetrically positioned and two test bearings at each
shaft end, see the basic design of Figure 10. The SKF-R2 test machines can
operate with radial, axial and combined load conditions. The load during testing is
hydraulically applied and monitored using load cells.
Each R2 test battery comprised 10 or 15 SKF-R2-rigs, capable of simultaneously
testing a bearing population of twenty to thirty bearing samples. All bearings of each
test battery are lubricated from one centralized oil lubrication system to achieve
homogeneous conditions for all test bearings. This lubrication system comprises
temperature conditioning and oil filtering and a large oil reservoir for up to 600 litres
of oil.
The control system of the lubricant flow, and of the temperature of the oil, can
provide specific lubrication conditions for the bearing under test, ranging from fully
flooded, thick oil film to reduced lubrication, down to very thin oil film boundary
conditions. Cleanliness of the lubricant is controlled by the high capacity multi-pass
filtration system down to 3 microns pore size.

Page 21 of 31
Figure 9 – Isometric view of SKF-R2 test rig

Figure 10 – Basic design of the SKF-R2 test rig.

Oil contamination levels during test can be monitored with an on-line automatic
particle counters also supported by microscopy and spectrographic analysis.

Post-test investigations of the failed bearings include the microscopic and

metrological examination of all failures with categorization of the failure type, origin,
and bearing component that is affected. Typically, all failures, either originated from
the surface, subsurface or from other critical components of the bearing, are
included in the fatigue life analysis,
The tenth percentile fatigue life of the tested bearings, and other statistical
parameters of the endurance test are evaluated performing the Weibull statistics
using a proprietary Weibull statistic software tool. The method of maximum
likelihood is applied to derive the experimentally observed L10 bearing life. The 90%

Page 22 of 31
confidence interval for the endurance tests is derived from fast Monte Carlo
simulations performed, by the software calculation tool, during the execution of the
Weibull analysis of the test data.

4.2 HYBRID BEARINGS FATIGUE LIFE TESTS

The hybrid bearing endurance tests reported here are related to bearing population
samples, that have been tested during the last twenty years. Typically, a test series
comprises 20 to 30 bearings, a total of twenty test series are included in this
evaluation amounting to a total of 509 hybrid bearings. These tests have produced a
total of 192 bearing failures of which only a very negligible number was related to
failures of the ceramic roiling element.
An overview of the different hybrid bearing types that were endurance tested and
the relevant information of the life tests, is given in Table 2. The correlation
between the observed tenth percentile fatigue life (i.e. L10,5 and L10,50) of the
tested bearings and the Generalized Bearing Life Model predictions is
presented in Figure 11.
Table 2 – Overview of the test conditions applied for hybrid bearing life testing.

Hybrid bearing Designation Test load Lubrication quality Contamination

type
Deep groove ball Clean: 0.8 ÷ 1
6305, 6309 1, 1.4, 2.1, 2.6 0.4, 1.5, 2.6, 3.3 Conta.: 0.02 ÷ 0.2
bearing
Angular contact Clean: 0.8 ÷1
7209, 7318 1.3, 1.8, 2.5, 2.9 0.1, 0.4,1.7, 2.6 Conta.: 0.02 ÷ 0.2
ball bearing
Cylindrical roller Clean: 0.8 ÷ 1
NU 209 1.7, 2.5, 3 0.4, 3.8 Conta.: 0.02 ÷ 0.2
bearing

A simple visual inspection of the test data presented in Figure 11 shows a good
degree of correlation between measured and calculated bearing life which indicates
the good ability of the model to predict hybrid bearing fatigue. Indeed, the Pearson
correlation coefficient between the GBLM predicted life and the experimental L10;50
observations is 0.91, i.e. 91%. This indicates a high degree of correlation between
the 10% bearing endurance and the GBLM life results.
The predicted fatigue lives are, in most cases, located within the endurance test
confidence interval of L10,5 and L10,50. As discussed earlier, this indicates an
adequate degree of experimental statistical significance for the GBLM model.
Furthermore, the coefficient of determination (R-square) of the GBLM model derived
from test data of Figure 11 is 98.6%, indicating a high degree of consistency
between the model prediction and the test data.

Page 23 of 31
Figure 11 - Generalized Bearing Life Model (GBLM) results plotted against L10 fatigue life observed in
endurance testing of hybrid bearings. Circular symbol indicates the median L10 life of the test, i.e. L10,50.
The vertical bar indicates the lower bound of the confidence interval of the test, i.e. L10,5.

It is to be pointed out that the experimental data displayed in Figure 11 are less in
number than the actual test series that were endurance tested. This is because,
through the years, some of the test were repeated under the same test conditions.
This allows the pooling of the results of two, or more test series, into one single
Weibull statistic. This method increases the precision of the statistical evaluation of
the life test results.
In few occasions hybrid and all-steel test series were endurance tested under similar
test conditions. Examples of this type of test results are given in Figure 12 and 13.
The results of two test series composed of thirty hybrids and standard all-steel
angular contact ball bearing are reported in Figure 12. The bearings were
endurance tested under high thrust load, clean and good lubrication.

Page 24 of 31
Figure 12 – Hybrid and all-steel bearings endurance tested under high load, good lubrication and clean
conditions. Calculated L10 are indicated with a circular symbol.

Figure 13 – Hybrid and all-steel bearings endurance tested under medium loading and severe
lubrication conditions. Calculated L10 are indicated with a circular symbol.

Under these running conditions subsurface Rolling Contact Fatigue will develop at a
faster rate than surface fatigue damage. The standard all-steel bearing which
operates at a lower contact pressure, has an advantage in terms of subsurface
fatigue life, as it is shown in Figure 12. The Generalized Bearing Life Model can
capture this type of behaviour well. The computed lives, indicated with a circular
symbol can consistently predict the tenth percentile fatigue life of the endurance
tests for both hybrid and the standard all-steel bearing variants.

An example of test results that are dominated by surface-initiated fatigue is

presented in Figure 13. The plot shows the Weibull statistic of the pooling of two test
series, hybrid and steel bearing variants. The pooled population sample was of forty-
five bearings for each test series which, (due to the severe contamination conditions

Page 25 of 31
in which the test was run), produced many failures. Therefore, the tenth percentile
fatigue life of the endurance test is obtained with good precision, i.e. narrow
confidence interval. In this case surface fatigue dominates the performance of the
bearing and the advantage, in terms of bearing life, goes in favour to the hybrid
bearing variant.

This is shown quite clearly by the test results presented in Figure 13 which indicates
a statistically significant advantage of hybrid bearing fatigue performance over the
all-steel counterpart. Also, in this case the predicted lives of the Generalized Bearing
Life Model are well aligned with the experimental observations.
Contrary to traditional Rolling Contact Fatigue that rely only on subsurface fatigue of
the rolling contact, the Generalized Bearing Life Model can account for both, surface
and subsurface fatigue damage. By computing the fatigue damage rate of the two
damaging processes, GBLM can recognize and account for the different fatigue
processes that are taking place in the rolling contact. This particularly, for the
conditions in which the better tribology and fatigue handling capabilities of the
ceramic-steel contact will compensate for the higher pressures present in the hybrid
contact. This new methodology provides realistic prediction of the expected fatigue
performance of hybrid bearings as shown in Figures 11, 12 ,13.

5 DISCUSSION
The use of hybrid bearings through the industry has developed quite steadily
through the years despite the higher cost involved in replacing traditional all-steel
with a hybrid bearing solution. Explanation of this market evolution can be found in
the ability of hybrid bearings to perform well in many difficult applications in which
all-steel bearings are found inadequate to satisfy the requirements.
This long bearing application experience is not always shared by bearing specialists
and researchers whose knowledge rely on traditional Rolling Contact Fatigue and
life testing that are typically performed under high contact pressures and good
lubrication conditions. These test conditions are unable to measure the actual
performance of hybrid bearings that are in use. Typical bearing applications are
dominated by surface-initiated fatigue. Traditional life test results are generally in
contrast with the performance of hybrid bearings used in the field. This issue has
puzzled researchers and engineers for years.
The present paper shows a novel approach in dealing with this long-standing
problem. This is achieved by developing a GBLM Index of surface fatigue stress for
hybrid bearings. This index is tailored to represent the tribology and fatigue
mechanisms of hybrid bearing raceway surfaces when the bearing is exposed to an
unfavourable and challenging lubrication environment.
The development of a GBLM fatigue damage model specific for the raceway surface
of hybrid bearings was helped by the research work of Brizmer [28] and Viellard [31].
Their work uncovered the mechanisms that distinguish the fatigue performance of a

Page 26 of 31
hybrid contact from an equivalent all-steel contact, particularly under conditions of
reduced lubrication and the presence of surface defects.
Roller bearings are in general dimensioned to withstand subsurface fatigue. Hence
subsurface initiated failures are normally avoided in most applications. Instead,
surface-initiated fatigue failures are still an issue in some applications.
The trend in the industry is to increase machine efficiency by reducing frictional
losses, (i.e. leading to thinner oil film), and by using higher speeds (i.e. leading to
higher temperatures and thinner oil film). The trend is that the environmental
conditions of rolling bearings tend to worsen throughout the years. In the meanwhile,
the service life and reliability requirements of mechanical systems have risen
following increased demand and tougher competition.
Considering all this, hybrid bearing special capabilities for increased resistance to
surface fatigue represent an important asset for today’s bearing applications. It is
expected that many modern industrial bearing applications will discover and take
advantage of the unique capabilities offered by hybrid ceramic bearings particularly
once that there is a better understanding of their performance.

6 CONCLUSIONS
Hybrid bearing fatigue life has been addressed by many authors and researchers.
Despite the great progress in fatigue performance of ceramics, no clear
performance advantage is recognized to hybrid bearings from the traditional Rolling
Contact Fatigue methodology used, up to now, in the bearing industry.
Rolling bearing fatigue life testing is in general performed under high loads and good
lubrication conditions which are of advantage for the RCF of standard all-steel
bearings. This because this type of life testing targets traditional subsurface fatigue
failure-mode. This behaviour is not reflected in the performance of hybrid bearings
as observed in field applications. Hybrid bearings are in general used for tough
lubrication conditions for which high surface fatigue resistance is required.
In many modern bearing applications, subsurface fatigue is negligible as it is
avoided by the proper selection of the bearing size. On the other hand, nowadays
surface fatigue caused by a harsh environment can be an issue in some bearing
applications. In this case, hybrid bearings have proven to perform better than all-
steel bearings.
A new approach to tackle RCF is proposed by the Generalized Bearing Life Model.
The new methodology can integrate hybrid bearing specific surface fatigue
properties into the calculation of the bearing life. In this way, GBLM can account for
the more favourable tribo-fatigue occurring in the ceramic-steel interface. The
ability of GBLM to forecast the tenth percentile fatigue life of hybrid bearings is good.
This was found after the examination of several endurance life tests of hybrid
bearings performed under a variety of loading, lubrication and contamination
conditions.

Page 27 of 31
The genialised bearing life methodology provides application engineers with a tool
for a realistic forecast of the expected performance of hybrid bearings. This allows a
proper account of the cost/benefit advantage of hybrid bearings and the further
growth of the use of this type of bearing throughout the industry.

References
[1] Zaretsky, E.V., Anderson, W.J., (1961). Rolling-contact fatigue studies with
four tool steels and a crystallized glass ceramic, J. Basic Eng, vol. 83, 603-612.
[2] Parker, R.J., Grisaffe, S.J., Zaretsky, E.V., (1965). Rolling-contact studies with
four refractory materials up to 2000° F, ASLE Trans., vol. 8, 208-216.
[3] Baumgartner, H.R., Evaluation of roller bearings containing hot pressed silicon
nitride rolling elements, (1973). Ceramics for High Performance Applications,
J.J. Burke, A.E. Gorum, and R.N. Katz, eds., Brook Hill Publishing Co.,
Chestnut Hill, Massachusetts, 713–727.
[4] Parker, R.J., Zaretsky, E.V., (1975), with discussion by Bernsby, R. M. and
Valori R. Fatigue Life of High Speed Ball Bearings with Silicon Nitride Balls,
Transactions of ASME, J. of Lubrication Technology, 97, 350-355.
[5] Delal, H.M., Chiu, Y.P., Rabinowicz, E., (1975). Evaluation of hot pressed
silicon nitride as a rolling bearing material, ASLE Trans., vol.18, 211–221.
[6] Lorösch, H. K., Vay, J., Weigand, R., Gugel, E., Kessel, H., (1980). Fatigue
Strength of silicon nitride for high-speed rolling bearings," Transactions of
ASME, J. of Engineering for Power, vol. 102, 128-131.
[7] Miner, J.R., Grace, W.A., Valori, R., (19810. A demonstration of high-speed
gas turbine bearings using silicon nitride rolling elements,” Lubr. Eng, vol. 37,
462-464, 473-478.
[8] Bhushan, B., Sibley, L.B., (1982). Silicon nitride rolling bearings for extreme
operating conditions”, ASLE Trans., vol. 25, 417- 428.
[9] Morrison, F.R., McCool, J.I., Yonushonis, T.M., Weinberg, P., (1984). The
Load-life relationship for M50 steel bearings with silicon nitride ceramic balls,
Lubr. Eng., vol. 40, 153 -159.
[10] Zaretsky, E.V., (1989). Discussion by Y.P. Chiu, T.E. Tallian, Ceramic
bearings for use in gas turbine engines, J. Materials Engineering, 11, 237-253.
[11] Zaretsky, E. V., Vlcek, B. L., Hendricks, R. C., (2005). Effect of Silicon Nitride
Balls and Rollers on Rolling Bearing Life, Tribology Transactions, 48:3, 425-
435
[12] Weibull, W., A Statistical Theory of Strength of Materials, (1939). Proc. of the
Royal Swedish Academy of Engineering Sciences, 151, 1–45,
[13] Lundberg, G., Palmgren, A., (1947). Dynamic Capacity of Rolling Bearings,
Acta Polytechnica, Mechanical Engineering Series, 1(3), 1–52.

Page 28 of 31
[14] Lundberg, G., Palmgren, A., (1952). Dynamic Capacity of Roller Bearings,
Acta Polytechnica, Mechanical Engineering Series, 2(4), 96–127,
[15] Ebert, F. J., (1990). Performance of silicon nitride (Si3N4) components in
aerospace bearing applications, ASME Publication No.90-GT-166.
[16] Cundill, R. T., (1990). Material selection and quality for ceramic rolling
elements. Proc. of Mech. Eng. Seminar, Rolling Element Bearings – Towards
the 21st Century, pp. 31-40.

[17] Cundill, R.T., (1993). High precision silicon nitride balls for bearings, Ball
Bearing Journal, 241, pp. 26-32.
[18] Galbato, A.T., Cundill, R.T., Harris, T.A., (1992). Fatigue Life of Silicon Nitride
Balls, Lubrication Eng. 48 (11), 886–894.
[19] Chiu, Y. P., Pearson, P. K., Dezzani, M. and Daverio, H., (1996). Fatigue Life
and Performance Testing of Hybrid Ceramic Ball Bearings, Lube Eng. 52, 3,
198-204.
[20] Dezzani, M. M., Pearson P. K. (1996). Hybrid Ceramic Bearings for Difficult
Applications. J. Eng. Gas Turbines Power, Vol. 118(2), 449-452.
[21] Forster, N. H., Peters, S. M., Chin, H. A., Poplawski, J. V., and Homan, R. J.,
(2017). “Applying Finite Element Analysis to Determine the Subsurface Stress
and Temperature Gradient in Highly Loaded Bearing Contacts,” Bearing Steel
Technologies: 11th Volume, ASTM STP1600, J. M. Beswick, Ed., ASTM,
West Conshohocken, PA, pp. 151–166
[22] Miner, J.R., Dell, J., Galbato, A.T., Ragen, M.A., (1996). F117-PW-100 Hybrid
Ball Bearing Ceramic Technology Insertion, Journal of Engineering for Gas
Turbines and Power, 118, pp. 434-442.
[23] Shoda, Y., Ijuin, S., Aramaki, H., Yui, H., and Toma, K. (1997). The
Performance of a Hybrid Ceramic Ball Bearing Under High Speed Conditions
with the Under-Race Lubrication Method, Tribol. Trans., 40, 4, pp.676-684.
[24] Jeng, Y.R., Huang, P.Y., (2000), Temperature rise of hybrid ceramic and steel
ball bearings with oil-mist lubrication. Lubrication Engineering, 56(12):18-24
[25] Wan, G.T.Y, Gabelli, A., Ioannides, E., (1997). Increased performance of
hybrid bearing with silicon nitride balls, Tribol. Trans. 40 (4) 701–707.
[26] Morales-Espejel, G.E., Gabelli, A., Vieillard, C., (2004). Hybrid bearings
lubricated with pure refrigerant liquid, Turbomachinery 32 (7) 387-394.
[27] Wallin, H.H., Morales-Espejel, G.E., (2011). Hybrid bearings in oil-free air
conditioning and refrigeration compressors, 〈http://evolution.skf.com〉.
Evolution Online.

Page 29 of 31
[28] Brizmer, V., Gabelli, A., Vieillard, C., Morales-Espejel, G.E., (2015). An
experimental and theoretical study of hybrid bearing micropitting performance
under reduced lubrication, Tribol. Trans. 58, 829–835.
[29] Morales-Espejel, G.E., Brizmer, V., (2011). Micropitting modelling in rolling-
sliding contacts: application to rolling bearings, Tribol. Trans. 54 625–643
[30] Hager, C. H., Jr., Doll, G. L., Evans, R. D., and Shiller, P. J. (2011). “Minimum
Quantity Lubrication of M50/M50 and M50/Si3N4 Tribological Interfaces, Wear,
271, 1761–1771.
[31] Vieillard, C., Kadin, Y., Morales-Espejel, G.E., Gabelli, A., (2016). An
experimental and theoretical study of surface Rolling Contact Fatigue damage
progression in hybrid bearings with artificial dents, Wear 364-365, 211–223
[32] Gabelli A., Morales-Espejel G.E., (2017). Improved Fatigue Life Analysis of
Pre-Dented Raceways Used in Bearing Material Testing”, Bearing Steel
Technologies 11th Volume, “Progress in Steel Technologies and Bearing
Steel Quality Assurance”, ASTM STP1600, 167–191
[33] Morales-Espejel, G. E., Gabelli, A. (2011). The Behaviour of Indentation Marks
in Rolling-Sliding Elastohydrodynamically Lubricated Contacts,” Tribol. T., Vol.
54, No. 4, pp. 589–606.
[34] Morales-Espejel, G.E., Gabelli, A., de Vries, A.J.C., (2015). A model for rolling
bearing life with surface and subsurface survival–tribological effects, Tribol.
Trans. 58, 894–906.
[35] Ioannides, E. and Harris, T.A., (1985). A New Life Model for Rolling Bearings,
Journal of Tribology, 107, 367–378,
[36] A. Gabelli, J. Lai, T. Lund, K. Ryden, I. Strandell, G.E. Morales-Espejel, (2012).
The fatigue limit of bearing steels. Part II: characterization for life rating
standards, Int. J. Fatigue 37, 155–168.
[37] Morales-Espejel, G.E., Gabelli, A., (2016). A model for rolling bearing life with
surface and subsurface survival: sporadic surface damage from deterministic
indentations, Tribology Int. 96, 279–288.

[38] Morales-Espejel, G. E., Wemekamp, A. W., Félix-Quiñonez, A., (2010),

“Micro-Geometry Effects on the Sliding Friction Transition in Elasto-
hydrodynamic Lubrication,” Proceeding of the Institution of Mechanical
Engineers - Part J. Journal of Engineering Tribology, 224, pp 621–637.
[39] Morales-Espejel, G. E., Gabelli, A., Ioannides, E., (2010), “Micro-Geometry
Lubrication and Life Ratings of Rolling Bearings,” Proceedings of the
Institution of Mechanical Engineers: Part C - Journal of Mechanical
Engineering Science, 224, pp 1965–1981, London, UK.
[40] Gabelli, A., Morales-Espejel, G. E., and Ioannides, E. (2008), Particle Damage
in Hertzian Contacts and Life Ratings of Rolling Bearings, Tribology
Transactions, 51, pp 428–445.

Page 30 of 31
[41] Goss, R., Ioannides, E., (1989), Performance testing facilities and
opportunities” Ball Bearing Journal, Special Issue, pp. 12 – 21.
[42] Rosado, L., Forster, N. H., Thompson, K. L., and Cooke, J. W., (2009) Rolling
Contact Fatigue Life and Spall Propagation Characteristics of AISI M50, M50 NiL,
and AISI 52100, Part I: Experimental Results, Trib. Trans., Vol. 53, No.1, pp. 29–41.

Page 31 of 31