Tribology International 119 (2018) 419–435

Contents lists available at ScienceDirect

Tribology International
journal homepage: www.elsevier.com/locate/triboint

Friction torque in rolling bearings lubricated with axle gear oils

Maroua Hammami a, c, *, Ramiro Martins a, Carlos Fernandes a, Jorge Seabra b,
Mohamed Slim Abbes c, Mohamed Haddar c
a
INEGI, Universidade do Porto, Campus FEUP, Rua Dr. Roberto Frias 400, 4200-465 Porto, Portugal
b
FEUP, Universidade do Porto, Rua Dr. Roberto Frias s/n, 4200-465 Porto, Portugal
c
Unit of Mechanics, Modelling and Manufacturing, National Engineers School of Sfax, University of Sfax, Sfax W-3038, Tunisia

A R T I C L E I N F O A B S T R A C T

Keywords: As a part of the main research project within the aim of increasing significantly drive axle efficiency, this work
Axle gear oils focuses on rolling bearing friction torque lubricated with five fully formulated axle gear oils with different vis-
Tribological behaviour cosity and different formulations. The lubricant tribological behaviour in different rolling bearings was analyzed.
Rolling bearings A modified Four-Ball Machine was used to test the rolling bearings. The effect of speed, temperature and axial
Friction torque load on rolling bearing friction torque was assessed. Experimental results for the internal friction torque were
Axle efficiency
validated with an SKF model. Direct comparisons in terms of friction torque between axle gear oils when they are
lubricating different rolling bearing types are presented and discussed.

1. Introduction
Power losses in automotive drive trains have become an important
field of investigation in automotive industry [1–5]. So far, the federal
standards and the state government regulations in vehicle fuel economy
combined with limits imposed by Environmental Protection Agency
(EPA) rules on CO2 emissions into the atmosphere have traditionally
been the main reason for this focus in drive train power losses [6–9]. The
looming energy crisis and increasing fuel prices have also added to the
motivation to reduce such losses.
It has been well settled that the automotive axle-differential is
considered as a significant contributor to power loss in the drive line
[10]. Axle efficiency values were reported to be as low as 90–95%
depending on the type of vehicle and on the applied torque and speed [2,
11,12]. Therefore, any tangible improvement to the axle efficiency has a
significant impact on the carbon emissions and the energy consumption
[3,11].
The axle transmission is a key component of the vehicle powertrain. It
is a very compact mechanical system, consisting generally of a hypoid
bevel geared transmission, tapered roller bearings, seals, shafts and an
axle gear oil [13]. The axle transmission requires a very high reliability,
since failures are not accepted by consumers [14–16].
As any power transmission system, axle power losses are divided into
several energy loss mechanisms as proposed in a number of papers by
Hohn et al. [17–19]. The sources of the major axle power losses were
dissipated mainly in friction loss between the meshing teeth of hypoid
gears [12,20–23], friction loss in the bearings [24–26], friction loss in the
seals [27] and spin losses due to lubricant pumping and churning, and
windage [5,28].
Axle power losses mainly originated in two type of losses. No-load
losses or churning losses generated by gears, rolling bearings and seals
that are related to the input speed, the operating conditions and mainly
related to the lubricant properties such as viscosity and density of the oil
[6–8,15] as well as immersion depth of the components in a sump
lubricated axle [29]. Rolling bearing losses depend on its type and size,
bearing arrangement, lubricant viscosity and supply [19]. These losses
are relevant but they are outside the scope of this research.
Load dependent losses generated by gears and rolling bearings are
related to the transmitted torque, coefficient of friction and sliding ve-
locity in the contact area of the components. Load dependent rolling
bearing losses also dependent on type and size, rolling and sliding con-
ditions and lubricant type [30]. In order to have a good prediction of the
power losses of the system, each component should be tested separately.
The rolling bearings are a major contributor to axle system power loss
[25] and their main function in axles is to support the pinion and the
differential gear under high load carrying capacity and high stiffness.
To achieve high efficiency in axle differentials, the reduction of in-
ternal friction torque in rolling bearings is of major concern. Thus, the
importance of understanding internal friction in rolling bearings be-
comes relevant. The energy saving and bearings performance

* Corresponding author. INEGI, Universidade do Porto, Campus FEUP, Rua Dr. Roberto Frias 400, 4200-465 Porto, Portugal.
E-mail address: mhammami@inegi.up.pt (M. Hammami).

https://doi.org/10.1016/j.triboint.2017.11.018
Received 20 August 2017; Received in revised form 6 November 2017; Accepted 11 November 2017
Available online 15 November 2017
0301-679X/© 2017 Elsevier Ltd. All rights reserved.
M. Hammami et al. Tribology International 119 (2018) 419–435

optimization are required [31].
The energy consumption due to rolling bearing power loss is
becoming more and more important when taking into account the focus
of science and industry on this issue.
Recently, automotive manufacturers and the rolling bearings manu-
facturers are trying to improve rolling bearing designs in order to reduce
the power loss generated, the energy consumption and the operating
temperatures and improve the lubrication conditions. At the same time,
they ask the lubricant manufacturers to provide new products that in-
crease rolling bearing life, while reducing the energy dissipated [25,26].
Only a limited number of studies focused on developing the axle
rolling bearings friction torque.
Spindler and von Petery (2003) [24] reported that INA (Industri-
e-NAdellager) has developed a new bearing design, where the tapered
roller bearings on the pinion shaft are replaced by double row
angular-contact ball bearing and the tapered roller bearings of the ring
gear shafts are replaced by single-row angular ball bearings. The benefits
of the rolling bearing substitution are no preload loss during operation
with 50% reduction of friction torque meeting the requirements for high
rigidity and long life.
Matsuyama et al. (2004) [25] published results when the super-low
friction torque tapered roller bearing supporting the pinion was used in
rear axle differentials. This study achieved a friction torque reduction of
80% compared to standard bearings.
Petery et al. (2004) [32] reported that INA and FAG in collaboration
with BMW conducted power loss measurements of the original bearing
design of a BMW axle with cross-locating taper roller bearing arrange-
ment and an alternative design with crosslocating double- and single-row
angular ball bearing arrangement. For medium load and speed and low
temperatures, relevant in the New European Driving Cycle (NEDC), the
bearing loss reduction for the alternative design was over 50 per
cent [19].
Hoshokawa et al. (2009) [26] proposed a new bearing concept which
is the double row angular contact ball bearing-so-called Tandem Ball
Bearings for rear axle drives. Through a comparative testing between a
new bearing design and standard bearing used in axles a relevant
reduction of 50% in friction torque can be achieved. This bearing concept
not only increases the service life but also make significant contribution
to lower fuel consumption by up 1.5% in every driving.
The motivation behind this study is conducting an accurate

Table 1
Axle gear oils properties.

Parameter Unit 75W85-B 75W90-A 75W90-B 80W90-A 75W140-A

candidate reference candidate reference reference

Base oil [
] PAO PAO PAO Mineral PAO

API/standard [
] – GL-4/GL-5/MT-1 – GL-4/GL-5/MT-1 GL-5

Chemical composition
Boron (B) [ppm] 0 – 81 – –
Calcium (Ca) [ppm] 1795 18 2891 97 33
Magnesium (Mg) [ppm] 6 1087 17 936 1093
Phosphorus (P) [ppm] 783 1622 958 1436 1686
Sulphur (S) [ppm] 2954 23262 3271 26947 22784
Zinc (Zn) [ppm] 899 7 1120 23 12

Physical properties
Density @ 15  C [g/cm3] 0.853 0.87 0.861 0.886 0.885
Thermal expansion coefficient [/]
8.1
7.3
7.6
7.7
6.8
(αt  10
4 )

Viscosity @ 40  C [cSt] 68.95 112.35 114.42 123.3 200.7

Viscosity @ 70  C [cSt] 23.86 36.7 38.14 34.86 61.86
Viscosity @ 100  C [cSt] 11.44 16.37 17.18 14.38 26.21

aA [/] 0.7
nA [/] 7.6655 7.5833 7.407 8.5027 7.1537
mA [/] 2.9663 2.9133 2.842 3.2783 2.7211

Thermoviscosity @ 40  C [K
1 ] 40.2 44.3 43.3 50.7 46.3

(β103 )
Thermoviscosity @ 70  C [K
1 ] 28.5 31.3 30.9 34.8 33.2
(β103 )
Thermoviscosity @ 100  C [K
1 ] 21.1 23.1 22.9 25 24.7
(β103 )

s @ 0.2 GPa [35] [/] 0.7382 0.7382 0.7382 0.9904 0.7382

t @ 0.2 GPa [35] [/] 0.1335 0.1335 0.1335 0.139 0.1335

Piezoviscosity @ 40  C [Pa
1 ] 1.291 1.387 1.39 1.934 1.498

(αGold  10
8 ) [35]
Piezoviscosity @ 70  C [Pa
1 ] 1.128 1.194 1.2 1.623 1.28
(αGold  10
8 ) [35]
Piezoviscosity @ 100  C [Pa
1 ] 1.022 1.072 1.079 1.435 1.142
(αGold  10
8 ) [35]

VI [/] 162 147 163 118 169

Tribofilm characterization
C 1s [At %] 38.87 48.83 35.01 39.11 53.79
Fe 2p [At %] 0.83 0.52 1.29 0.39 –
O 1s [At %] 40.37 40.47 44.24 47.41 37.07
Ca 2p [At %] 7.68 – 10.23 0.39 –
Mg 1s [At %] – 2.58 – 3.88 2.24
P 2p [At %] 8.75 7.15 5.91 7.58 6.38
S 2p [At %] 1.21 0.45 2.01 0.6 0.5
Zn 2p3 [At %] 2.3 – 1.32 0.07 –

420
M. Hammami et al.
measurement of internal friction torque of different rolling bearing types
which is challenging since experimental measurements must be deter-
mined precisely. A modified Four-Ball Machine (Cameron-Plint TE 82/
7752) was suited. The tests were performed under a wide range of
operating conditions (load, speed, temperature and lubricant).
Then, the power loss of rolling bearings can be predicted using
models based on a large number of measurements and they have been
developed by bearing manufacturers such as SKF and FAG. For this work,
the model selected to analyze the results obtained is the SKF Friction
Torque Model [33]. This model will be calibrated using the experimental
results from tests performed with the selected rolling bearing types,
lubricated with several axle gear oils having different formulations. A
better understanding of the influence of the oil's formulation in the
rolling bearing's power loss after the calibration of the model is expected.
The present work provides new knowledge about torque loss in
different rolling bearings types lubricated with axle gear oils. Extensive
tests were performed and a considerable amount of experimental results
of power loss in rolling bearings, difficult to find in literature, were ob-
tained. The model calibration allows a better understanding of the in-
fluence of the rolling bearing geometry and the oil's formulation in the
bearing's power loss for different loads and speeds. The rolling bearing
power loss model will be relevant for the global axle differential power
loss model predictions.
2. Axle gear oils properties
Five multigrade axle gear oils, available on the market and with large
range of kinematic viscosity, were selected. All the lubricants are fully
synthetic oils except for the 80W90-A product which is a semi synthetic
oil. Three among them (75W90-A, 80W90-A and 75W140-A), are refer-
ence (A) oils and labelled as “Fuel Efficient”, and the other two products
(75W85-B and 75W90-B) are candidate (B) oils, as presented in Table 1.
Several experimental analyses were performed to characterize the
chemical, physical, rheological and tribological properties of the selected
axle oil formulations and the measured properties are displayed in
Table 1. A detailed presentation of those analysis and properties
Fig. 1. Schematic view of the rolling bearing assembly.
421
Tribology International 119 (2018) 419–435
measured can be found in Ref. [34].
3. Materials and methods
3.1. Rolling bearing assembly
Rolling bearing tests were performed on a modified Four-Ball Ma-
chine where the four-ball arrangement was replaced by a rolling bearing
assembly as shown in Fig. 1. This assembly allows to test several rolling
bearings and to obtain reliable friction torque measurements at different
operating temperature. Additional details of this assembly can be found
in Ref. [31].
The rolling bearing assembly is composed mainly of two parts. The
first part is bearing lower race (3) which is fitted in the spacer (2) and this
set is fitted on the bearing housing (1). These parts of the group (A) are
tight clamped to ensure that there is no relative motion between them.
The second part is the bearing upper race (5) is mounted on the shaft
adapter (6).
In operation, the load (P) is applied on the lower plate (13) and the
rotational speed (n) is transmitted to the shaft adapter (6), which is
connected to the drive shaft of the machine (see Fig. 1).
In order to measure the internal bearing torque, a piezoelectric torque
cell KISTLER 9339A was used, ensuring high accuracy measurements (∓
1 Nmm) even when the friction torque generated in the bearings was very
low compared to the measurement range available.
During the test, the temperature at different points is recorded. Five
thermocouples (I-V) are positioned in strategic locations in order to
measure the lubricant and bearing housing temperatures. This assembly
is also monitored with two other thermocouples used to measure the
temperatures of the room and of the air flowing around the
bearing housing.
The rolling bearing assembly is submitted to continuous forced air
convection by two fans, having 38 mm in diameter and running at
2000 rpm, evacuating the heat generated during bearing operation. Also,
to control the temperature during the tests, the bearing assembly is
mounted with two heaters which are controlled with a PID control
M. Hammami et al. Tribology International 119 (2018) 419–435

Table 2
Characteristics of TBB 51107, RTB 81107 TN and TRB 320/28 X/Q.

Thrust Roller Taper roller

ball thrust bearing
bearing bearing TRB 320/28
TBB RTB 81107 X/Q
51107 TN

Principal d mm 35 28
dimensions H mm 12 16
D mm 52

Basic load Dynamic C kN 19.9 29 31.9

ratings Static C0 kN 51 93 38

Speed ratings Reference rpm 5600 2800 9500

speed
Limiting rpm 7500 5600 13000
Fig. 2. Rolling bearing housing with heaters controlled by a PID system. speed

system with feedback given by thermocouple III (see Fig. 2). The control
SKF), the Cylindrical Roller Thrust Bearing (RTB, ref. 81107 SKF) and the
system can assure a temperature variation always lower than ∓ 1  C. The
Taper Roller Bearing (TRB, ref. 320/28 X/Q. SKF). The dimensions and
rolling bearing assembly presented above is suitable for ball and roller
characteristics of the selected geometries are reported in Fig. 3
bearings whose dimensions are limited by the bearing housing and the
and Table 2.
machine itself.

3.2. Rolling bearings tested 3.3. Operating conditions

A wide range of rolling bearing geometries might be used in axle
differentials. Generally, the axle contains eight bearings. Two taper
rolling bearings that support the input pinion and two more to support
the hypoid gear wheel. The two half shafts are also supported by two
rolling bearings. Given that, these outer bearings are physically remote
from the main drive unit they are considered thermally distinct and their
contribution was neglected [28].
The modified Four-Ball Machine permits the friction torque mea-
surements with specific types of rolling bearings whose dimensions are
limited by a maximum outer diameter of 56 mm and a maximum width
of 16 mm.
With such limitations, it is crucial to understand the behaviour of
different rolling element bearings, understand the lubrication capabil-
ities of different axle gear oils and try to reduce the power loss in rolling
bearings, knowing the influence of several parameters such as speed,
load, oil formulation and bearing geometry.
The experimental tests were performed with three types of rolling
bearing geometries, lubricated with the selected axle gear oils. The
rolling bearings selected are the Thrust Ball Bearing (TBB, ref. 51107
The oil level should reach the centre of the lowest rolling element as
indicated by the manufacturer [33] and the oil volume required is
approximately 14 ml for TBB and RTB and 8 ml for TRB as indicted by
Cousseau et al. [31]. A new rolling bearing is mounted for each oil test, in
order to avoid the possible chemical interactions between oils and even
to reduce the influence of the surface finish.
The surface roughness of all rolling bearing raceways was measured,
using an absolute stylus probe in a HommelWereke T4000 device. The
measurements confirm that the surface roughness of all raceways tested
with the different axle gear oils was very similar.
The test machine allows a maximum inner raceway speed of
1500 rpm. However, above 1200 rpm all axle gear oils promote full-film
lubrication regime. Once this condition is reached the EHL coefficient of
friction is defined. Measurements above 1200 rpm will not provide
additional data useful for the torque loss model of rolling bearings. For
that, the tests were performed for different rotational speed between 75
and 1200 rpm, allowing to cover all lubrication regimes, from boundary
to full-film lubrication.
The friction torque was measured for different rotational speeds, in

Fig. 3. TBB 51107, RTB 81107 TN and TRB 320/28 X/Q geometries.

422
M. Hammami et al. Tribology International 119 (2018) 419–435

Table 3 Table 5
Ball-raceway contact parameters for TBB 51107 rolling bearing. Roller-raceway contact parameters for TRB 320/28 X/Q rolling bearing.

Unit Raceway Ball Unit Raceway Roller

RXi [m] ∞ 310
3 RXi [m] 17.2610
3 2.9510
3

RYi [m]
3.3810
3 310
3 RX [m] 5.910
3
RX [m] 610
3 l [m] 12.1610
3
RY [m] 53.410
3 σc [μm] 0.25
σc [μm] 0.18
Axial load [N] 4000 7000
Axial load [N] 4000 7000
a [μm] 41.79 55.28
a [μm] 102.8 123 Ac [mm2] 1.016 1.345
Ac [mm2] 0.14 0.198 p0 [GPa] 0.957 1.266
p0 [GPa] 2.06 2.47
δ [μm] 5.22 7.48
3.4. Test procedure

Table 4
Before starting each test, a running-in period is always required for
Roller-raceway contact parameters for RTB 81107 TN rolling bearing.
each rolling bearing and is carried under an axial load of 1000 N and
Unit Raceway Roller
increasing rotational speed from 75 to 1200 rpm during 10 min.
RXi [m] ∞ 2.510
3 For each test with the selected rolling bearing, the axial load is
RX [m] 510
3 applied and the rotational speed set to the required value and the oper-
l [m] 510
3
ating temperature is imposed. When the machine reaches the stabilized
σc [μm] 0.14
temperature (thermal equilibrium) due to the heaters, four friction tor-
Axial load [N] 4000 7000
que measurements were performed. Three values were kept and the most
a [μm] 33.22 43.5 dispersed one was disregarded. The torque measurement should be made
Ac [mm2] 0.332 0.435 in short periods of time (120 s) at stabilized temperatures (±2  C) in
p0 [GPa] 0.766 1.004
order to avoid the drift effect of the torque cell [31]. Measurements
during longer periods (above 120 s) show a decay of the applied torque
the range 75 up to 1200 rpm. This speed range is within the limits of the due to temperature gradients and are intrinsic to piezo-electric torque
test machine, allowing to cover all lubrication regimes, from boundary to cells. So, the friction torque value (for each rotational speed and load) is
full-film lubrication. the average value of the three closest measurements.
According to the axle lubricant temperatures measured during an EPA
(Environmental Protection Agency) driving cycle including both city and
highway cycles, the operating temperature of 70  C was selected [36]. 3.5. Taper rolling bearings - assembly, operating conditions and test
All rolling bearings were submitted to an axial load of 7000 N, as- procedure
suring a sufficiently high bearing rigidity [24,26]. The characteristics of
the contact between the rolling elements (ball or roller) and the raceway Several modifications were introduced into the test assembly, the
were displayed in Tables 3 and 4 for TBB 51107 and RTB 81107, operating conditions and the test procedure in order to measure the in-
respectively. For TBB and RTB the normal load is equal to the applied ternal friction torque in Taper Rolling Bearings (TRB). A different shaft
axial load. adapter (see Fig. 4), minimizing the misalignment of the TRB, and a
lower race support were manufactured. Instead of a cone Morse
connection, the superior part of the shaft adapter is a hollow shaft design
to assemble on the machine input shaft. Furthermore, both pieces have a
hole across their sections and perpendicular to their rotational axles with
a screw applied to prevent the sliding effect among the two parts.
The characteristics of the contact between the rollers and the raceway

Fig. 4. Schematic view of the TRB assembly. Fig. 5. Simplified inter geometry for TRB [37].

423
M. Hammami et al. Tribology International 119 (2018) 419–435

Fig. 6. Temperature variation (θ) of tapered roller bearings during the running-in period (t).

are displayed in Table 5 for TRB 320/28 X/Q. For the TRB the normal Fa =sinðψÞ
contact load between the roller and raceway is given by Equation (1)
Fn ¼ (1)
z
according to Eschamann [37] (see Fig. 5).
where

Fig. 7. Viscosity ratio for TBB, RTB and TRB under 70  C and 7000 N operating conditions.

424
M. Hammami et al.
Fig. 8. Drag losses for different rolling bearing geometries TBB 51107 and RTB 81107
under lubricated with 75W90-A at 70  C.
e
ψ ¼ arctan
1:5
It was necessary to take into account that tapered roller bearings
should have a long running-in period, which is characterized by a sig-
nificant amount of friction between the rolling elements and the race-
ways, and it can be identified by the temperature spike shown in Fig. 6.
After reaching the maximum operating temperature, a gradual decrease
of the temperature is observed until a stabilized value is achieved. The
time necessary to reach the stabilization temperature depends on the
lubricant formulation and, consequently, the running-in time is not pre-
defined. Instead, the running-in period is complete when the temperature
variation is lower than 1  C per hour.
4. Film thickness inside rolling bearings
The film thickness (h0 ) of the rolling element/raceway contact is
calculated using the Hamrock and Dowson [38] equation (see Appendix
A) for elliptical contacts (TBB) and Dowson and Higginson [39] equation
(see B) for line contacts (RTB and TRB).
The corrected film thickness (h0C ) is given in Equation (A.6) which is
the product of the central film thickness with the thermal correction
factor (ϕT ), as shown in Equation (A.7).
The concept of specific film thickness (Λ) is usually used to determine
the lubrication regime. The specific film thickness is calculated with
Equation (3).
h0c
Λ¼
σc
The composite roughness σ c is calculated through the roughness of
the rolling element (σ element , ball or roller depending on bearing geometry)
and the roughness of the race (σ race ), according to Equation (4). The
roughness values for each rolling bearing tested are presented
in Tables 3–5.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
σc ¼ ðσ element Þ2 þ ðσ race Þ2
To evaluate the film thickness and lubrication regime in rolling
bearings, the concept of viscosity ratio may also be used to evaluate lu-
bricant's effectiveness in forming sufficient lubricant film to separate the
surfaces in contact. The viscosity ratio is calculated using Equation (5).
ν two contributions: the rolling and sliding torques, respectively, M 0 rr and
k¼
ν1
This parameter depends essentially on the mean diameter dm of the
rolling bearing without selecting the type of rolling bearing. For that
Tribology International 119 (2018) 419–435
reason, the viscosity ratio can be related to the specific film thickness, as
it is shown in Equation (6).
k ffi Λ1:3 (6)
Fig. 7 (a) presents the viscosity ratio, calculated using the SKF cata-
logue [33], for TBB under constant temperature of 70  C and a load of
7000 N. Using this concept it is possible to observe that the viscosity ratio
increases when the speed increases. The five axle oils are classified ac-
cording to their viscosity at the operating temperature. The 75W140-A
oil has the highest viscosity generating the highest viscosity ratio while
75W85-B oil has the lowest viscosity and promotes the lowest viscosity
ratio. The viscosity ratio values of oils 75W90-B, 75W90-A and 80W90-A
are placed between the previous two, depending directly on
their viscosity.
In general [33], a rolling bearing operates under boundary film
regime for a viscosity ratio 0.4 < k < 1, under mixed-film lubrication
regime for 1.0 < k < 2.0 and under full-film lubrication regime for 2.0
< k < 4.0.
From 75 rpm up-to 1200 rpm, and for all axle gear oil formulations,
boundary and mixed-film lubrication are the dominant lubrication re-
(2) gimes. At 1200 rpm full-film lubrication regime is achieved for all lu-
bricants except for the 75W85-B oil.
Fig. 7 (b) shows the variation of viscosity ratio in roller-raceway
contact of RTB with the rotational speed under the applied loads of
7000 N and at 70  C. It is possible to observe that the viscosity ratio
increases when the speed increases up to 600 rpm. Axle gear oils share a
similar behaviour as TBB tests for the speed range before 600 rpm since
the temperature is maintained constant. The 75W85-B oil with the lowest
viscosity and piezo-viscosity had an opposite behaviour of the 75W140-A
oil, promoting the lowest viscosity ratio. The viscosity ratio values of oils
75W90-B, 75W90-A and 80W90-A are placed between the previous two,
depending directly on their kinematic viscosity (see Table 1).
Fig. 7 (c) displays the viscosity ratio values for TRB under the same
operating conditions. It was observed that the lubricants were ranked
with a similar trend as the viscosity ratio values in TBB tests. Generally,
the viscosity ratio values in TRB are slightly lower than the viscosity ratio
values in TBB. For the given range of rotational speeds and for all the axle
oil formulations, boundary and mixed film lubrication are the dominant
lubrication regimes (k  2), except at 1200 rpm.
5. Friction torque model
In order to understand the experimental results as well as to predict
the power loss in rolling bearings lubricated with axle gear oils, a friction
torque model is required. For that SKF proposed a detailed model [33]
which divides the total friction torque in its true physical components. It
takes into account four different torque losses as shown in the
(3)
following equation:
Mt ¼ M'rr þ Msl þ Mseal þ Mdrag (7)
Since the rolling bearings tested do not have seals, the Mseal compo-
nent was disregarded in the calculation.
The drag losses are very small because the operating speeds and the
mean diameter of the rolling bearings are also small (dm ¼ 43.5 mm for
TBB and RTB), consequently, the drag torque loss term was also
(4)
disregarded.
Fig. 8 presents the usual drag torque loss for two different rolling
bearings, respectively TBB 51107 and RTB 81107. The results show that
the drag torque losses are negligible for these rolling bearings, when
compared with the other torque loss mechanisms.
Therefore, the total friction torque of the rolling bearings had only
(5)
Msl , as presented in the following equation:
Mt ¼ M'rr þ Msl (8)
425
M. Hammami et al. Tribology International 119 (2018) 419–435

Fig. 9. Results of TBB 51107 lubricated with axle gear oils at constant temperature of 70  C under an axial load of 7000 N.

Assuming that the friction torque obtained from experimental mea- All the following Equations (10)–(15) are necessary to calculate the
surements was equal to the total torque loss predicted by the SKF model rolling and sliding torques.
(Mt ¼ Mtexp ) and the rolling torque was accurately calculated, the sliding h i
torque can be directly determined (see Equation (9)). M 0 rr ¼ ϕish ⋅ϕrs ⋅ Grr ⋅ðn⋅υÞ0:6 (10)

Msl ¼ Mt
M'rr ¼ Mtexp
M'rr (9)

426
M. Hammami et al. Tribology International 119 (2018) 419–435

constant temperature of 70  C and an axial load of 7000 N. Fig. 9 (a)

shows that, in general, the measured total friction torque of the TBB
decreases when the operating speed increases from 75 rpm to 1200 rpm,
except in the case of oil 75W140-A, for which the friction torque
increased as the speed increases. It is clear that, for low rotational speeds,
the two candidate (B) oils generated higher values of the total friction
torque than the reference (A) oils (see Fig. 9 (a)) due to the significant
differences between them in terms of the additive packages present in
their formulations.
Fig. 9 (b) shows that when the speed increased the specific film
thickness inside the TBB increased from 0.22/0.46 up to 1.30/2.75,
depending on the tested oil, meaning that boundary and mixed lubrica-
tion regimes prevail under these operating conditions as indicated by
Fig. 9 (f) as well.
The results of the rolling friction torque inside the thrust ball bearing
are displayed in Fig. 9 (c). As expected, at constant temperature (70  C)
the rolling torque increases when the speed increases (M'rr ∝ðn⋅υÞ0:6 ).
Comparing the behaviour of the different oil formulations, the 75W140-
Fig. 10. Temperature evolution under constant temperature measurements of 70  C for A oil, with the highest operating viscosity, generated the highest rolling
RTB 81107. torques, while 75W85-B oil, with the lowest viscosity, generated the
lowest rolling torques. The rolling torques of the other three oil formu-
1 lations follow the trend of their viscosities. So the main parameter that
ϕish ¼ (11)
1 þ 1:84⋅10
9 ⋅ðn⋅dm Þ1:28 ⋅ν0:64 differentiates the lubricants in the calculation of the rolling torque is the
viscosity at the operating temperature.
1 As expected the sliding torque, presented in Fig. 9 (d), was higher
ϕrs ¼ pffiffiffiffiffiffiffiffiffi
Kz
(12) than the rolling torque, presented in Fig. 9 (c). The sliding torque curves
Krs ⋅ν⋅n⋅ðdþDÞ⋅ 2⋅ðD
dÞ
e present the same trend of the total friction torque curves.
In fact, under constant operating temperature, the film thickness in-
Msl ¼ Gsl ⋅μsl (13) creases as the operating speed increases, the specific film thickness also
increases thus generating lower sliding torque loss, as shown in Fig. 9 (b).
μsl ¼ ϕbl ⋅μbl þ ð1
ϕbl Þ⋅μEHL (14) In a real application, without temperature control, when the speed
increases the temperature also increases. Higher speeds promote higher
1 film thickness and lower torque loss, while higher temperature promotes
ϕbl ¼ 1:4 (15) a reduction of oil viscosity, and consequently a decrease of film thickness
e2:6⋅10
8 ⋅ðn⋅νÞ ⋅dm
and an increase on sliding torque loss. However, a real application under
The equations related to the geometry of each rolling bearings used full-film lubrication and without temperature control tends to reach a
are presented in Appendix C. constant specific film thickness no matter the rotational speed [40].
The rolling torque (Equation (10)) depends on the bearing type, These counter effects are in general in favour of speed increase up to a
rotational speed, oil viscosity and two factors, namely the inlet shear certain thermal limit, above which, there will be no benefits in increasing
heating ϕish , and the kinematic replenishment reduction factor ϕrs . the operating speed. This is well demonstrated by the inlet shear heating
The sliding torque (Equation (13)) was highly influenced by the effect on film thickness.
bearing type, the coefficient of friction and the lubrication regime. The The experimental sliding torques can be used to determine the
lubrication regime on the model is quantified by the ϕbl quantity where experimental sliding coefficient of friction, using Equation (16), as pre-
the coefficient of friction under full film (μEHL ) and boundary (μbl ) sented in Fig. 9 (e). The sliding coefficient of friction μsl follows the trend
lubrication were presented. of the sliding torque.
Knowing the value of the sliding friction torque determined using Fig. 9 (f) shows the weighting factor ϕbl , which is always between
Equation (13), it is possible to calculate the sliding coefficient of friction 0 and 1. This parameter decreases when the speed increases. The 75W85-
using Equation (16). The sliding coefficient of friction (μexp sl ) is now B presents the highest values while the 75W140-A presents the lowest
considered as an experimental coefficient of friction determined through values. Other formulations have values in between. The weighting factor
rolling bearing tests. is inversely proportional to the viscosity of the lubricant, explaining this
behaviour. The weighting factor presents a direct influence on the sliding
Msl Mtexp
M'rr
μexp
sl ¼ ¼ (16) coefficient of friction μsl presented in Fig. 9 (e). It should also be noticed
Gsl Gsl
that the values of ϕbl for each oil formulation almost cover the whole
Furthermore, the sliding coefficient of friction can be effectively range of values (0 ϕbl 1), meaning that the operating conditions
predicted by this model using Equation (14) once the boundary film represent all the lubrication regimes, from boundary to full-film
coefficient of friction-μbl and the full-film coefficient of friction-μEHL lubrication.
values are known.
The coefficients of friction μbl and μEHL are calculated minimizing the 6.2. Cylindrical roller thrust bearings experimental results (RTB 81107)
difference between μsl and μexp
sl through Equations (14) and (16).
The experimental tests with cylindrical roller thrust bearings, lubri-
6. Experimental and model results cated with axle gear oils, were carried out under the same constant
operating temperature (70  C) and a high axial load (7000 N). However,
6.1. Thrust ball bearings experimental results (TBB 51107) it was not possible to keep a constant operating temperature of 70  C
above 600 rpm, as shown in Fig. 10. In those cases, it was necessary to
The tests of TBB lubricated with axle oils were carried out under a cool the bearing housing, and that was not achieved. At 1200 rpm, the
maximum temperature reached 135  C with 75W90-A oil.

427
M. Hammami et al.
Fig. 11. Results of RTB 81107 lubricated with axle gear oils at constant temperature of 70  C with an axial load of 7000 N.
Fig. 11 (a) displays the total friction torque measured for RTB in all
operating conditions. The total friction torque decreases with increasing
speed, up to 600 rpm and T ¼ 70  C. Above 600 rpm, it is interesting to
notice that the reference (A) oils generated significantly higher operating
temperatures than the candidate (B) oils, whose operating temperature
remained constant (70  C) at 900 rpm and reached 85  C at 1200 rpm.
Up to 600 rpm, 80W90-A oil generally produced the lowest values of
the total friction torque while 75W90-A oil usually generated the highest
corresponding values. Oils 75W140-A, 75W85-B and 75W90-B presented
intermediate values. Above 600 rpm the candidate (B) oils generated the
428
Tribology International 119 (2018) 419–435
lowest friction torques.
Fig. 11 (b) shows that when the operating speed increases from
75 rpm to 600 rpm the specific film thickness inside the RTB increases
from 0.21 to 0.45 up to 0.87–1.81, meaning that mixed film lubrication
prevails. All axle gear oils share the same trend, where 75W140-A oil
produced the highest Λ because of its high viscosity at 70  C and 75W85-
B oil generated the lowest Λ because of its low viscosity. Above 600 rpm
the specific film thickness is not constant and strongly dependent on the
test temperature.
The rolling torque calculated for RTB under the operating conditions
M. Hammami et al.
Table 6
Values of the coefficients μbl and μEHL for TBB 51107 and RTB 81107, under 7 kN and at
70  C.
Valid for 3262:5 < n⋅dm < 52200
Bearing type
Lubricant Parameter TBB RTB
75W85-B and 75W90-B μbl 0.124 0.073
μEHL 0.056 0.032
75W90-A and 80W90-A μbl 0.097 0.071
μEHL 0.064 0.036
75W140-A μbl 0.078 0.073
μEHL 0.063 0.040
is presented in Fig. 11 (c). It is observed that when the speed increases the
rolling torque increases. This figure also shows that oils 75W90-B,
75W90-A and 80W90-A generated very similar rolling torques, because
they have similar viscosities at 70  C.
Equation (9) was used to obtain the sliding friction torque (see Fig. 11
(d)). This figure shows that the sliding torque decreases with the increase
of speed for all operating conditions. The lubricant behaviour presents
the same trend of their total friction torque. However, it is not possible to
compare the sliding torques of the different axle gear oils, below and
above 600 rpm, because of the large differences in operating
temperatures.
The sliding coefficients of friction are presented in Fig. 11 (e),
showing that reference (A) oils 80W90-A and 75W140-A have the lowest
sliding torques below 600 rpm.
The weighting factor ϕbl , shown in Fig. 11 (f), does explain the
behaviour of the sliding coefficient of friction μsl up to 600 rpm. Above
600 rpm, the ϕbl values are not comparable.
6.3. Comparison between TBB and RTB
It is interesting to compare the friction behaviours of the TBB and RTB
rolling bearings. Under constant temperature (70  C) and constant load
(7000 N), the total friction torque decreases when the speed increases,
both for TBB and RTB. However, the maximum torques measured for
each type of bearing at 75 rpm are 370 N.mm and 1060 N.mm for TBB
and RTB, respectively, showing the major role and influence of the
rolling bearing type and geometry on the total friction torque Mt .
When the speed increases, the rolling torque, Mrr , shows a small in-
crease in the case of TBB and a large increase in the case of RTB (see Fig. 9
(c) and 11 (c)). The maximum rolling torques calculated for each type of
bearing are 92 N.mm for TBB and 142 N.mm for RTB, showing again the
influence of the rolling bearing geometry on the rolling torque.
The reduction of the sliding torque, Msl , with increasing speeds, is
much more significant in a RTB than in a TBB, as may be noticed
comparing Fig. 9 (d) and 11 (d).
These differences in the internal friction torque (Mt , Mrr and Msl )
between TBB and RTB are related to the geometry of the contact area, an
elliptical contact in the case of TBB and a line (rectangular) contact in the
case of the RTB, and to the sliding speed inside the contact, which is
higher in the case of the RTB. These differences are reflected also on the
values of (Grr and Gsl ), which are significantly higher in the case of RTB.
As mentioned in section 5, the minimization of the difference be-
tween experimental and model values of the sliding coefficient of friction
allows the calculation of the μbl and μEHL values. In this case, the values of
μbl and μEHL were determined for each type of rolling bearing (TBB and
RTB) and for each type of axle gear oil formulation according to three
groups (group I: 75W85-B and 75W90-B, group II: 75W90-A and 80W90-
A and group III: 75W140-A), as presented in Table 6.
The boundary coefficient of friction, μbl , depends on the gear oil
formulation and rolling bearing type, as shown in Table 6. For all the
lubricant formulations (A and B), the boundary coefficient of friction μTBBbl
is larger than μRTBbl . In the case of the TBB, μbl for the candidate (B) oils
429
Tribology International 119 (2018) 419–435
reached a high value (μTBB bl ¼ 0.124), meaning that the formulations
containing Zinc generate very high sliding torque under boundary film
lubrication. This behaviour was not observed with RTB.
The full-film coefficient of friction, μEHL , also depends on the gear oil
formulation and rolling bearing type. The reference and the candidate
oils have similar behaviours when lubricating the two rolling bearing
EHL > μEHL ). In general, the candidate (B) oils, which have lower
types (μTBB RTB
viscosities also generated lower μEHL coefficients.
The Stribeck curves of the five axle gear oils lubricating the two
rolling bearing types are presented in Figs. 12 and 13, for TBB and RTB,
respectively. Those figures present the experimental results with the
error bar for each value shown with markers and the model simulations
shown by the continuous lines, in function of the modified Hersey
parameter Sp ¼ Ur⋅η⋅α0:5 ⋅Fn
0:5 .
In general, the approximation of the sliding coefficient of friction μsl ,
predicted by the model, is quite good, whatever the axle gear oil
formulation and rolling bearing type. It is also clear that there is a better
approximation at high speeds and high viscosities (larger values of Sp). At
low speed, under boundary film lubrication conditions, the scatter of the
μsl values is larger.
6.4. Tapered roller bearings experimental results (TRB 320/28 X/Q)
The results of the total friction torque measurements carried out on
TRB, under a 7 kN axial load and at 70  C, are presented in Fig. 14. Mt
decreases with the increase of the rotational speed, while the corre-
sponding specific film thickness increased from 0.22 to 0.45 up to
1.29–2.46, that is from boundary film lubrication up to mixed film
lubrication, (see Fig. 14 (b)).
At low speeds (n < ¼ 300 rpm) the candidate (B) formulations
generated higher friction torques than the standard (A) formulations. In
fact, Mt75W90
B is 1800 N.mm while Mt75W140
A reaches 760 N.mm, that is
2.4 times higher. At 600 rpm and above, all axle gear oils generated very
similar friction torques.
Fig. 14 (a) shows that the internal friction torque of the TRB MtTRB
decreases drastically between 75 and 300 rpm, and above 300 rpm re-
mains almost constant and very similar for all axle gear oil formulations.
In fact, above 300 rpm MtTRB is significantly lower than MtRTB . This
behaviour and the corresponding measurements of the internal friction
torque for the TRB are probably related to the fact that the axial load
applied generates a very small normal load between the rollers and the
raceways, due to the internal geometry of the 320/28 X/Q tapered roller
bearing. Unfortunately, the modified Four-Ball Machine is not prepared
for axial loads above 7 kN. Furthermore, it is possible that the friction
torque model used (see section 5) is not prepared nor calibrated to deal
with this type of loading (pure axial load and low normal load).
These measurements of the total friction torque carried out on TRB,
are only indicative and eventually as a first approximation of the friction
torque behaviour of the different axle gear oil formulations in a TRB.
7. Conclusions
Three rolling bearing types (TBB, RTB and TRB) were tested in a
modified Four-Ball Machine, using five different axle gear oil formula-
tions. The internal friction torque was measured for operating speeds
between 75 and 1200 rpm. The friction torque measurements were
correlated using a rolling bearing friction torque model.
 The experimental results show that the rolling bearing geometry has a
prominent influence on the internal friction torque of the rolling
bearings and on the corresponding torque and power loss. Among the
three types of rolling bearings, TBB presented the lowest friction
torque loss, followed by RTB and TRB (MtTBB < MtRTB < MtTRB ). At low
speed (75 rpm) and for a reference axle gear oil formulation (75W90-
A), the torque loss on TBB, RTB and TRB are 290, 1060 and
M. Hammami et al. Tribology International 119 (2018) 419–435

Fig. 12. Model simulations for sliding coefficient of friction against modified Hersey Fig. 13. Model simulations for sliding coefficient of friction against modified Hersey
parameter for the TBB 51107 lubricated with axle gear oils at constant temperature of parameter for the RTB 81107 lubricated with axle gear oils at constant temperature of
70  C with an axial load of 7000 N. 70  C with an axial load of 7000 N.

430
M. Hammami et al.
Fig. 14. Results of TRB 320/28 X/Q lubricated with axle gear oils at constant temperature of 70  C with an axial load of 7000 N.
1200 N.mm, respectively. This means that the elliptical contacts in-
side the TBB generate significantly less torque loss than the line
contacts of the RTB and TRB, and that substantial gains in efficiency
might be obtained replacing roller bearings by ball bearings. In fact, it
was not possible to run the RTB tests at constant temperature above
600 rpm, because the corresponding torque and power loss was too
high, which did not occur in the case of the TBB.
 The five axle gear oils tested showed substantial differences in torque
loss, mainly under boundary to mixed film lubrication (specific film
thickness Λ below 1.0 and weighting factor ϕbl greater than 0.5).
Under these conditions, the low viscosity candidate (B) oils (75W85-B
and 75W90-B) generated higher torque loss than the reference (A)
oils (75W90-A, 80W90-a and 75W140-A), for all types of rolling
bearings, although these differences were smaller in the case of the
Cylindrical Roller Thrust Bearings (RTB). Under full-film lubrication
(specific film thickness Λ over 1.0 and weighting factor ϕbl lower than
0.5), the differences in torque loss among the five axle gear oils are
small and they only reflect the differences in viscosity and in rolling
torque of these oil formulations.
 It is assumed that the rolling torque Mrr only depends on the operating
speed and kinematic viscosity, besides the rolling bearing geometry.
Thus, oils 75W85-B and 75W140-A, generated the lowest and highest
rolling torques, respectively, because they also have the lowest and
the highest kinematic viscosities. The other axle gear oil formulations
(75W90-A, 75W90-B and 80W90-A) have very similar viscosities and
generated very similar rolling torques (in between the other two
formulations), which is consistent with the experimental measure-
ments of the torque loss.
 The sliding torque Msl was determined using Equation (9). The
analysis of Fig. 9 (d) and 11 (d) show that, with increasing speeds the
differences in torque loss among the five axle oil formulations
decrease, becoming almost null for the TBB tested at 1200 rpm. This is
not so clear in the case of the RTB, because above 600 rpm it was not
possible to keep the operating temperature constant (70  C). The
sliding torque loss (Msl ) results confirm that, under boundary to
mixed film lubrication (λ below 1.0 and ϕbl greater than 0.5), the low
viscosity candidate (B) oils (75W85-B and 75W90-B) generated
higher sliding torque loss than the reference (A) oils (75W90-A,
80W90-a and 75W140-A).
Appendix A. Equations for centre film thickness in point or elliptical contact
The central film thickness in the ball-raceway contact of the TBB is calculated using Equation (A.1) [38].
431
Tribology International 119 (2018) 419–435
 The analysis of Fig. 9 (f) and 11 (f) show that the range of operating
speeds allowed to cover the full range of the weighting factor ϕbl ,
almost from 1.0 down to 0.0. This means that all lubrication regimes
were considered, from boundary film lubrication, through mixed film
regime, till full-film lubrication, for all oil formulations and both
rolling bearing types. Thus, the calculation of coefficient of sliding
friction also covers all lubrication regimes and the values of μbl and
μEHL , shown in Table 6, can be considered as representative of the axle
gear oils tested. It is clear that the candidate (B) formulations have
higher values of the coefficient of boundary friction than the standard
(A) formulations (μBsl > μAsl ), thus generating higher torque loss values
at low speed. It is also clear that the candidate (B) formulations have
lower values of the full-film coefficient of friction than the standard
(A) formulations (μBEHL < μAEHL ), thus generating lower torque loss
values at high speed, mainly because they have lower kinematic
viscosities.
 Three comparisons deserve special attention. Formulations 75W90-A
and 75W90-B have very similar physical properties, but have
different additive packages and generate different tribo-films [34].
However, the differences in friction torque loss between these two
formulations are not very significant, both for TBB and RTB, although
in favour of the 75W90-A. Formulations 75W90-A and 80W90-A,
have the same additive package but different base oils, PAO vs min-
eral, respectively. However, they generate almost the same friction
torque loss, whatever the operating conditions, justifying the perti-
nence of the mineral oil formulation for some applications. In fact,
these two formulations have exactly the same values for the boundary
and full-film coefficients of friction, that is μ75W90
A ¼ μ80W90
A and
bl bl
μEHL
75W90
A
¼ μEHL
80W 90
A
. Finally, as expected, the heavy duty formula-
tion 75W140-A exhibits a quite low friction torque loss under
boundary film lubrication conditions.
Acknowledgements
The authors gratefully acknowledge the funding of Project NORTE-
01-0145-FEDER-000022 - SciTech - Science and Technology for
Competitive and Sustainable Industries, cofinanced by Programa Oper-
acional Regional do Norte (NORTE2020), through Fundo Europeu de
Desenvolvimento Regional (FEDER).
M. Hammami et al. Tribology International 119 (2018) 419–435

h0 ¼ 1:345⋅RX ⋅U 0:67 ⋅G0:53 ⋅W
0:067 ⋅C0 (A.1)

Equations (A.2) to (A.5) define all the parameters which are needed to calculate the central film thickness in TBB.

η0 ⋅ðU1 þ U2 Þ
U¼ (A.2)
2⋅RX ⋅E

G ¼ 2⋅α⋅E (A.3)

2⋅Fn
W¼ (A.4)
E ⋅R2X

 0:64
RY

0:752 RX
C0 ¼ 1
0:61⋅e (A.5)
Those parameters depend mainly on the geometry of the ball–raceway contact in the TBB as well as the axle gear oil properties at the corresponding
operating temperatures, 70  C for this case of study.
The theoretical film thickness, defined by Equation (A.1), was corrected using the thermal reduction factor ϕT where the inlet shear heating of the
lubricant was considered, as shown in Equations (A.6)-(A.9).
Then, the specific film thickness defined as the ratio of centre lubricant film thickness and the composite root mean square surface roughness of the
thrust ball bearing (σ c ¼0.18 μm) was computed using Equation (4). The ball and raceway surface roughness measurements were made in previous work
with an absolute stylus probe in a Hommelwereke T4000 device. The absolute stylus probe measures both the waviness and roughness of the surface
while the differential probe only measures the surface roughness.
To evaluate the corresponding lubrication regime, the specific film thickness in the ball–raceway contact should be calculated.

h0C ¼ ϕT ⋅h0 (A.6)

   
1
ϕT ¼ 1 þ 0:1 1 þ 14:8⋅Ve0:83 ⋅L0:64 (A.7)

jU1
U2 j
Ve ¼ (A.8)
U1 þ U2

β⋅η⋅ðU1 þ U2 Þ2
L¼ (A.9)
KL

Appendix B. Equations for centre film thickness in line or rectangular contact

The central film thickness in the roller-raceway contact of the RTB and TRB is calculated using Equation (B.1) [41].

h0 ¼ 0:975⋅RX ⋅U 0:727 ⋅G0:727 ⋅W
0:091 (B.1)

Equations from (B.2) to (B.4) define all the parameters which are needed to calculate the central film thickness in RTB and TRB.

η0 ⋅ðU1 þ U2 Þ
U¼ (B.2)
2⋅RX ⋅E

G ¼ 2⋅α⋅E (B.3)

Fn
W¼ (B.4)
RX ⋅l⋅E 
Those parameters depend mainly on the geometry of the roller–raceway contact in the RTB and TRB as well as the axle gear oil properties at the
corresponding operating temperatures, 70  C for this case of study.
The theoretical film thickness, defined by Equation (B.1), was corrected using the thermal reduction factor ϕT where the inlet shear heating of the
lubricant was considered, as shown in Equations (B.5)-(B.8).
Then, the specific film thickness defined as the ratio of central lubricant film thickness and the composite root mean square surface roughness (for
cylindrical roller thrust bearing σ c ¼ 0.14 μm and for tapered roller bearings σ c ¼ 0.25 μm) was computed using Equation (4). The roller and raceway
surface roughness measurements were made in previous work with an absolute stylus probe in a Hommelwereke T4000 device (measuring length
4.8 mm, cut-off 0.8 mm). To evaluate the corresponding lubrication regime, the specific film thickness in the ball–raceway contact should be calculated.

h0C ¼ ϕT ⋅h0 (B.5)

   
1
ϕT ¼ 1 þ 0:1 1 þ 14:8⋅Ve0:83 ⋅L0:64 (B.6)

432
M. Hammami et al. Tribology International 119 (2018) 419–435

jU1
U2 j
Ve ¼ (B.7)
ðU1 þ U2 Þ

β⋅η⋅ðU1 þ U2 Þ2
L¼ (B.8)
KL

Appendix C. SKF friction torque model

Appendix C.1. Thrust ball bearing (TBB)

Grr ¼ R1⋅dm1:83 ⋅Fa0:54 (C.1)

Gsl ¼ S1⋅dm0:05 ⋅Fa4=3 (C.2)

Appendix C.2. Cylindrical roller thrust bearing (RTB)

Grr ¼ R1⋅dm2:38 ⋅Fa0:31 (C.3)

Gsl ¼ S1⋅dm0:62 ⋅Fa (C.4)

Appendix C.3. Tapered roller bearing (TRB)

Grr ¼ R1⋅dm2:38 ðFr þ R2⋅Y⋅Fa Þ0:31 (C.5)

Gsl ¼ S1⋅dm0:82 ðFr þ S2⋅Y⋅Fa Þ (C.6)

The model constants for each rolling bearing geometry are presented in Table C.7.

Table C.7
SKF friction torque model constants for each rolling bearing geometry.

Constants TBB RTB TRB

R1 1.0310
6 2.2510
6 2:38  10
6

R2 – – 10.9
R3 – – –
S1 1.610
2 0.154 0.014
S2 – – 2
Kz 3.8 4.4 6
Krs 310
8

Notation and units

a radius of the contact circle [m]

aA D341 viscosity parameter [
]
Ac area of contact [mm2]
C basic dynamic load rating capacity [N]
C0 basic static load rating [N]
C0 ellipticity parameter [
]
d bearing bore diameter [mm]
D bearing outside diameter [mm]
dm rolling bearing mean diameter [mm]
e rolling bearing SKF calculation factor [
]
E contact equivalent Young modulus [Pa]
Fa axial load [N]
Fn normal load [N]
Fr radial load [N]
G material parameter [
]
Grr factor depending on the bearing type, bearing mean diameter and applied load [N mm]
Gsl factor depending on the bearing type, bearing mean diameter and applied load [N mm]
H width of the bearing [mm]
h0 centre film thickness [m]
h0C corrected centre film thickness [m]
k viscosity ratio [
]

433
M. Hammami et al. Tribology International 119 (2018) 419–435

KL thermal conductivity [W/mK]

Krs starvation constant for oil bath lubrication [
]
Kz bearing type related geometry constant [
]
l roller element width [m]
L thermal parameter of the lubricant [
]
mA D341 viscosity parameter [
]
Mdrag friction torque of drag losses [Nmm]
M'rr rolling friction torque [Nmm]
Mseal friction torque of seals [Nmm]
Msl sliding friction torque [Nmm]
Mt internal bearing friction torque [Nmm]
Mtexp total bearing friction torque measured experimentally [Nmm]
n rotational speed [rpm]
nA D341 viscosity parameter [
]
p0 maximum Hertz pressure [Pa]
R1; 2; 3 geometry constant for rolling bearing friction torque [
]
RX equivalent radius of curvature in x direction [m]
RXi curvature radius of element i in x direction [m]
RY equivalent radius of curvature in y direction [m]
RYi curvature radius of element i in y direction [m]
s pressure-viscosity parameter [
]
S1; 2 geometry constant for sliding friction torque [
]
t pressure-viscosity parameter [
]
U speed influence parameter [
]
U1 linear speed of ball [m/s]
U2 linear speed of ring [m/s]
Ur rolling speed [m/s]
Ve sliding factor of the roller/ball [
]
VI Viscosity Index [
]
W load parameter [
]
Y rolling bearing SKF calculation factor [
]
z number of rollers [
]
α pressure-viscosity coefficient [Pa
1 ]
αt thermal expansion coefficient [
]
β thermoviscosity coefficient [K
1 ]
η0 dynamic viscosity [Pa s]
ϕbl sliding friction torque weighting factor [
]
ϕish inlet hear heating reduction factor [
]
ϕrs kinematic replenishment/starvation reduction factor [
]
ϕT thermal reduction factor [
]
ψ contact angle [rad]
Λ specific film thickness [μm]
μbl coefficient of friction in boundary film lubrication [
]
μEHL coefficient of friction in full film lubrication [
]
μsl coefficient of sliding friction [
]
ν kinematic viscosity [cSt]
ν1 required kinematic viscosity [cSt]
σc composite roughness [m]
δ penetration [μm]

References [6] Bala V, Rollin AJ, Brandt G. Rheological properties affecting the fuel economy of
multigrade automotive gear lubricants. 2000. Tech. rep., SAE Technical Paper,
https://doi.org/10.4271/2000-01-2051.
[1] Joachim FJ, Kurz N, B€ orner J. Reduction of power losses in transmissions and
[7] Bala V, Brandt G, Walters D. Fuel economy of multigrade gear lubricants. Industrial
gearings. Appl Mech Mater 2011;86:883–8. https://doi.org/10.4028/www.
Lubr Tribol 2000;52(4):165–73. arXiv, https://doi.org/10.1108/
scientific.net/AMM.86.883.
00368790010333610. https://doi.org/10.1108/00368790010333610.
[2] Joachim F, B€ orner J, Kurz N. Power losses in transmissions, axles, and steering
[8] Bjornen KK, Chambers H, DeGonia D. Development of a fuel efficient multipurpose
systems. In: Encyclopedia of lubricants and lubrication. Springer; 2014.
75w-90 gear lubricant. 2003. Tech. rep., SAE Technical Paper, https://doi.org/10.
p. 1398–411.
4271/2003-01-1992.
[3] Xu H, Singh A, Kahraman A, Hurley J, Shon S. Effects of bearing preload, oil
[9] Porrett DC, Miles SD, Werderits EF, Powell DL. Development of a laboratory axle
volume, and operating temperature on axle power losses. J Mech Des 2012;134(5):
efficiency test. 1980. Tech. rep., SAE Technical Paper, https://doi.org/10.4271/
054501. https://doi.org/10.1115/1.4006325.
800804.
[4] Katoh H, Inoh T, Umezawa K. Exact measurement of power loss in automotive
[10] Hurley JD. An experimental investigation of thermal behavior of an automotive rear
transmission and axle for fuel economy. 1983. Tech. rep., SAE Technical Paper,
axle. Ph.D. thesis. The Ohio State University; 2009.
https://doi.org/10.4271/830573.
[11] Talbot D, Kahraman A, Li S, Singh A, Xu H. Development and validation of an
[5] Ko Y, Hosoi K. Measurements of power losses in automobile drive train. 1984. Tech.
automotive axle power loss model. Tribol Trans 2016:1–44. just-accepted, https://
rep., SAE Technical Paper, https://doi.org/10.4271/840054.
doi.org/10.1080/10402004.2015.1110862.

434
M. Hammami et al.
[12] Kolivand M, Li S, Kahraman A. Prediction of mechanical gear mesh efficiency of
hypoid gear pairs. Mech Mach Theory 2010;45(11):1568–82. https://doi.org/10.
1016/j.mechmachtheory.2010.06.015.
[13] Kolekar AS. Lubrication & efficiency of rear wheel drive axles in road vehicles.
Ph.D. thesis. 2013.
[14] Evans, SD. Delivering axle efficiency and fuel economy through optimised fluid
design, SAE Technical PaperHttp://dx.doi.org/10.4271/2014-01-2799.
[15] Johnston G, Wayte R, Spikes H. The measurement and study of very thin lubricant
films in concentrated contacts. Tribol Trans 1991;34(2):187–94. https://doi.org/
10.1080/10402009108982026.
[16] Guangteng G, Cann P, Olver A, Spikes H. Lubricant film thickness in rough surface,
mixed elastohydrodynamic contact. J Tribol 2000;122(1):65–76. https://doi.org/
10.1115/1.555330.
[17] Hohn, BR, Michaelis, K, Vollmer, T, Thermal rating of gear drives: balance between
power loss and heat dissipation, AGMA Technical Paper.
[18] H€ohn B-R, Michaelis K, Hinterstoißer M. Optimization of gearbox efficiency. Goriva
i maziva 2009;48(4):462.
[19] Seabra J, Michaelis K, H€ ohn B-R, Hinterstoißer M. Influence factors on gearbox
power loss. Industrial Lubr Tribol 2011;63(1):46–55.
[20] Cheng Y, Lim TC. Vibration analysis of hypoid transmissions applying an exact
geometry-based gear mesh theory. J Sound Vib 2001;240(3):519–43. https://doi.
org/10.1006/jsvi.2000.3247.
[21] Cheng Y, Lim TC. Dynamics of hypoid gear transmission with nonlinear time-
varying mesh characteristics. J Mech Des 2003;125(2):373–82.
[22] Wang J, Lim TC, Li M. Dynamics of a hypoid gear pair considering the effects of
time-varying mesh parameters and backlash nonlinearity. J Sound Vib 2007;308(1):
302–29. https://doi.org/10.1016/j.jsv.2007.07.042.
[23] Xu H, Kahraman A. Prediction of friction-related power losses of hypoid gear pairs.
Proc. Inst. Mech. Eng, Part K. J Multi-body Dyn 2007;221(3):387–400. https://doi.
org/10.1243/14644193JMBD48.
[24] Spindler D, Von Petery G. Angular contact ball bearings for a rear axle differential.
2003. Tech. rep., SAE Technical Paper, https://doi.org/10.4271/2003-01-3743.
[25] Matsuyama H, Dodoro H, Ogino K, Ohshima H, Toda K. Development of super-low
friction torque tapered roller bearing for improved fuel efficiency. 2004. Tech. rep.,
SAE Technical Paper, https://doi.org/10.4271/2004-01-2674.
435
Tribology International 119 (2018) 419–435
[26] Hosokawa RS, de Oliveira RAA, Franco DM, de Aguiar Vendrasco A. Double row
regular contact ball bearing for axle drives. 2009. Tech. rep., SAE Technical Paper,
https://doi.org/10.4271/2009-36-0055.
[27] Lin H, Burke DC, Binoniemi RR, Wenstrup L, Woodard T. Seal friction effect on
drive axle efficiency. 2005. Tech. rep., SAE Technical Paper, https://doi.org/10.
4271/2005-01-3779.
[28] Kolekar AS, Olver AV, Sworski AE, Lockwood FE. The efficiency of a hypoid axle a
thermally coupled lubrication model. Tribol Int 2013;59(0):203–9. https://doi.org/
10.1016/j.triboint.2012.03.013.
[29] Changenet C, Velex P. Housing influence on churning losses in geared
transmissions. J Mech Des 2008;130(6):062603. https://doi.org/10.1115/1.
2900714.
[30] Fernandes CM, Marques PM, Martins RC, Seabra JH. Gearbox power loss. part i:
losses in rolling bearings. Tribol Int 2015;88:298–308. https://doi.org/10.1016/j.
triboint.2014.11.017.
[31] Cousseau T, Graça B, Campos A, Seabra J. Experimental measuring procedure for
the friction torque in rolling bearings. Lubr Sci 2010;22(4):133–47. https://doi.
org/10.1002/ls.115.
[32] von Petery G. Fuel economy by custom-made: bearings for differentials from bmw.
ATZ Worldw 2004;106(12):11–2.
[33] SKF. Skf rolling bearings general catalogue. 2013.
[34] Hammami M, Rodrigues N, Fernandes C, Martins R, Seabra J, Abbes MS, et al. Axle
gear oils: friction, wear and tribofilm generation under boundary lubrication
regime. Tribol Int 2017. https://doi.org/10.1016/j.triboint.2017.04.018.
[35] Gold, JW, Schmidt, A, dicke, H, Loos, H, Aßmann, C.. Viscosity-pressure-
temperature behaviour of mineral and synthetic oils, J Synthetic Lubr 18(1).
[36] Hammami M, Martins R, Abbes MS, Haddar M, Seabra J. Axle gear oils: tribological
characterization under full film lubrication. Tribol Int 2017;106:109–22. https://
doi.org/10.1016/j.triboint.2016.05.051.
[37] Eschmann P. Ball and roller bearings - theory, design and applications. 1985.
[38] Hamrock BJ, Dowson D. Ball bearing lubrication. John Wiley & Sons; 1981. p. 386.
[39] Dowson D, Higginson GR. Elastohydrodynamic lubrication. S. I. Editon. Pergamon
Press Ltd; 1977.
[40] Fernandes CM, Martins RC, Seabra JH. Friction torque of thrust ball bearings
lubricated with wind turbine gear oils. Tribol Int 2013;58:47–54.
[41] Dowson D, Higginson G. A numerical solution to the elasto-hydrodynamic problem.
J Mech Eng Sci 1959;1(1):6–15.