sustainability

Article
The Bearing Stiffness Effect on In-Wheel Motors
Matej Biček 1 , Raphaël Connes 1 , Senad Omerović 1 , Aydin Gündüz 2 , Robert Kunc 3, * and
Samo Zupan 3
1 Elaphe Ltd., Teslova 30, Ljubljana 1000, Slovenia; matej.bicek@elaphe-ev.com (M.B.);
raphael.connes@elaphe-ev.com (R.C.); senad.omerovic@elaphe-ev.com (S.O.)
2 Turkish Aerospace Inc., Ankara 06980, Turkey; dr.aydingunduz@gmail.com
3 Faculty of Mechanical Engineering, University of Ljubljana, Ljubljana 1000, Slovenia; samo.zupan@fs.uni-lj.si
* Correspondence: robert.kunc@fs.uni-lj.si; Tel.: +386-1-477-1508




Received: 3 April 2020; Accepted: 12 May 2020; Published: 15 May 2020


Abstract: In-wheel motors offer a promising solution for novel drivetrain architectures of future electric
vehicles that could penetrate into the automotive industry by transferring the drive directly inside the
wheels. The available literature mainly deals with the optimization of electromagnetically active parts;
however, the mechanical design of electromagnetically passive parts that indirectly influence motor
performance also require detailed analysis and extensive validation. To meet the optimal performance
of an in-wheel motor, the mechanical design requires optimization of housing elements, thermal
management, mechanical tolerancing and hub bearing selection. All of the mentioned factors have an
indirect influence on the electromagnetic performance of the IWM and sustainability; therefore, the
following paper identifies the hub bearing as a critical component for the in-wheel motor application.
Acting loads are reviewed and their effect on component deformation is studied via analytically and
numerically determined stiffness as well as later validated by measurements on the component and
assembly level to ensure deformation envelope and functionality within a wide range of operations.

Keywords: air-gap; hub bearing; in-wheel motor; mathematical stiffness model; validation tests

1. Introduction
Conventional mobility with Internal Combustion Engines (ICE) and complex drivetrains as shown
on Figure 1 is facing competition with electric vehicles within the passenger car market and lately also
within commercial segments. Showcasing electric vehicles is a thing of the past as they are pushed by
incentives and national
Sustainability or PEER
2020, 12, x FOR associative
REVIEW directives to being frequently driven on global2 of
roads.
18

1. Internal
Figure Figure Combustion
1. Internal Engines(ICE)
Combustion Engines (ICE) drivetrain
drivetrain with
with an an all-wheel
all-wheel drive withdrive with obvious
obvious
complexity
complexity [1]. [1].

The automotive evolution includes charging infrastructure establishment, energy source

development and propulsion architecture selection. In-Wheel Motor (IWM) platforms (Figure 2)
intrinsically allow higher design space, lowering of the vehicle’s center of gravity, reduction of
Sustainability 12, 4070;
2020, parts
required doi:10.3390/su12104070
for vehicle www.mdpi.com/journal/sustainability
propulsion [2,3], and consequentially offers cost reduction potential [4,5].
Higher energy efficiency and increased range [6–8] are also met as no mechanical transmission is
required and the wheels are propelled directly [4,9,10]. More space for passengers and cargo [10–13]
allows chassis designers to utterly change the way cars look and perform [14,15] with components
Sustainability 2020, 12, 4070 2 of 18

Figure 1. Internal Combustion Engines (ICE) drivetrain with an all-wheel drive with obvious
complexity [1].
The automotive evolution includes charging infrastructure establishment, energy source
development The and propulsion
automotive architecture
evolution selection.
includes charging In-Wheel establishment,
infrastructure Motor (IWM)energy platforms
source(Figure 2)
intrinsically allow and
development higher design space,
propulsion lowering
architecture of the
selection. vehicle’s
In-Wheel center
Motor of gravity,
(IWM) reduction
platforms (Figure of 2) required
parts for vehicle allow
intrinsically propulsion [2,3], and
higher design consequentially
space, lowering of the offers cost
vehicle’s reduction
center potential
of gravity, reduction[4,5].
of Higher
energyrequired partsand
efficiency for increased
vehicle propulsion [2,3], are
range [6–8] and also
consequentially
met as no offers cost reduction
mechanical potentialis[4,5].
transmission required and
Higher energy efficiency and increased range [6–8] are also met as no mechanical transmission is
the wheels are propelled directly [4,9,10]. More space for passengers and cargo [10–13] allows chassis
required and the wheels are propelled directly [4,9,10]. More space for passengers and cargo [10–13]
designers to utterly change the way cars look and perform [14,15] with components not needing fixed
allows chassis designers to utterly change the way cars look and perform [14,15] with components
mechanical coupling
not needing fixed allowing
mechanicalfree arrangement
coupling in the
allowing free vehicle. in the vehicle.
arrangement

FigureFigure 2. In-Wheel
2. In-Wheel Motor(IWM)
Motor (IWM) propulsion
propulsion platform
platformwith Elaphe
with M700M700
Elaphe all-wheel drive showing
all-wheel drive showing
simplicity [1].
simplicity [1].
The reason for not having this technology already adopted in the automotive world likely stems
The reason for not having this technology already adopted in the automotive world likely stems
from the lack of required know-how for the design of high torque density, innovations in high
from the lack ofsealing
diameter required withknow-how for the design
low loss generation, noise,ofvibration,
high torqueand density,
harshnessinnovations in high diameter
(NVH), lightweight
sealingdesign,
with low withloss
stiffgeneration,
and durablenoise, vibration,
housing and harshness
components. The most(NVH),
commonlightweight
Permanent design,
Magnet with stiff
and durable
Synchronoushousing components.
Motor–PMSM The most
for in-wheel commonhas
applications Permanent Magnet
an outer rotor with Synchronous Motor–PMSM
a relatively small air
gap between rotor and stator [8]. Identification of mechanical failure modes and effects
for in-wheel applications has an outer rotor with a relatively small air gap between rotor and stator [8]. have been
presented in a review paper [13] with keywords such as unsprung mass, eccentricity, moment of
Identification of mechanical failure modes and effects have been presented in a review paper [13] with
inertia, vibrations and hub bearing faults. The latter is rated as a component with one of the highest
keywords such as unsprung mass, eccentricity, moment of inertia, vibrations and hub bearing faults.
Risk Priority Numbers (RPN) and this paper describes in detail the reasoning behind it.
The latter is rated as a component with one of the highest Risk Priority Numbers (RPN) and this paper
describes in detail
Sustainability 2020, 12,the reasoning
x FOR behind it.
PEER REVIEW 3 of 18
The following chapters deal with the identification of bearing function, loads acting on it,
The following chapters deal with the identification of bearing function, loads acting on it,
stiffness evaluation with numerical and experimental approach as well as validation on bearings and
stiffness evaluation with numerical and experimental approach as well as validation on bearings and
assembled IWM.
assembled IWM.
2. Bearing Function in IWM
2. Bearing Function in IWM
Integration ofof
Integration a hub bearing
a hub unit
bearing into
unit thethe
into in-wheel motor
in-wheel motoris is
objected
objectedtoto
ensure thethe
ensure rotation whereas
rotation
offering required stiffness to support axial and vertical loads as shown in Figure 3.
whereas offering required stiffness to support axial and vertical loads as shown in Figure 3.

Figure 3. Axial and radial loads acting on rear wheels of a vehicle.

Figure 3. Axial and radial loads acting on rear wheels of a vehicle.

Vertical loads are directed opposite to gravitational force, whereas axial loads occur in the
cornering direction. The bearing is most affected by the bending moment resulting from the lateral
(cornering) force, defined as MX-Y, which acts via pneumatic tire’s effective rolling radius and rim, as
shown in Figure 4. FY is shown as an example for left cornering shown as FY-L in Figure 3.
Sustainability 2020, 12, 4070 3 of 18

Figure 3. Axial and radial loads acting on rear wheels of a vehicle.

Vertical loads
Vertical areare
loads directed opposite
directed to gravitational
opposite to gravitational force, whereas
force, whereasaxialaxialloads
loadsoccur
occurininthe cornering
the
direction. The bearing is most affected by the bending moment resulting from
cornering direction. The bearing is most affected by the bending moment resulting from the lateral the lateral (cornering)
force,
(cornering) as Mdefined
definedforce, X-Y , which
as Macts via pneumatic
X-Y, which tire’s effective
acts via pneumatic rolling
tire’s effective radius
rolling andand
radius rim, asasshown in
rim,
4. FinY Figure
shown
Figure 4. FYas
is shown is an
shown as an example
example for left cornering
for left cornering shown as FY-L
shown as Fin in Figure
Y-LFigure 3. 3.

Figure 4. Bending moments generated by the axial and radial force.

Figure 4. Bending moments generated by the axial and radial force.
Moment MX-Z is much smaller since it is resulting from the vertical force and the small distance
Moment MX-Z is much smaller since it is resulting from the vertical force and the small distance
from the tire center to the point of rotational deflection. As identified, the most critical loads are severe
from the tire
braking, center to
cornering andthe pointover
driving of rotational deflection. As
a road pothole/obstacle identified,
causing the load.
an impact mostMcritical
X-Z from
loads are
severe braking,
vertical cornering
impacts and driving
or MX-Y from over a road
severe cornering pothole/obstacle
can reach causing
values that result an deflection
in large impact load. MX-Z from
angles
vertical impacts
and should beor MX-Y from
anticipated severe
during thecornering canHub
design stage. reach values
bearing that result
deflection in large
is less deflection
problematic for angles
andconventional vehicle corners,
should be anticipated where
during thethedesign
deflection actsHub
stage. on the movement
bearing of discisbrake
deflection towards
less problematic for
braking pads inside the caliper (Figure 5). The objective of every brake manufacturer
conventional vehicle corners, where the deflection acts on the movement of disc brake towards braking is to design a
padsbraking
insideassembly, which
the caliper will be5).
(Figure functional and notof
The objective affect wheel
every rotation
brake during severe
manufacturer is cornering.
to design a braking
assembly, which will be functional and not affect wheel rotation during severe cornering.
Sustainability 2020, 12, x FOR PEER REVIEW 4 of 18

Figure
Figure 5. 5.Conventional
Conventional suspension
suspension with
withcomponents
components in display. Hub Hub
in display. bearing deflection
bearing angle βangle
deflection is β is
limited
limited bybybraking
brakingpads
pads inside
inside the
thecaliper
caliperand
andcan be be
can larger in comparison
larger to when
in comparison integrated
to when inside inside
integrated
an an IWM.
IWM.

Bearingmanufacturers
Bearing manufacturers forfor this
thisreason
reasondevelop different
develop solutions
different to reduce
solutions the brake
to reduce pedal pedal
the brake
travel and improve brake caliper piston knock-back [16]. In case of too large hub bearing deflection,
travel and improve brake caliper piston knock-back [16]. In case of too large hub bearing deflection,
brakes can also endure non-uniform wear of rotor and brake lining which has a negative effect on
brakes can also endure non-uniform wear of rotor and brake lining which has a negative effect on
vehicle handling performance. The air gap between rotor and stator of the most common IWM layout
vehicle handling
is shown performance.
in Figure The air
6 as the envelope gap between
in which rotordeflection
hub bearing and stator of the most
is allowed. common
The gap is alsoIWM
used layout
is shown
to coverinelastic
Figuredeformations
6 as the envelope
of motor in housing
which hub
due bearing deflection
to accelerations, is allowed. Theloads,
thermomechanical gap is also
used to coverand
production elastic deformations
assembly tolerances,ofdeformations
motor housing
due due to accelerations,
to press fits and residual thermomechanical
stresses resulting loads,
production and assembly
from manufacturing tolerances, deformations due to press fits and residual stresses resulting from
processes.
manufacturing processes.
 Figure
travel 5. Conventional
and improve suspension
brake caliper pistonwith components
knock-back incase
[16]. In display.
of tooHub
largebearing deflection
hub bearing angle β is
deflection,
limited
brakes can by braking
also endurepads inside thewear
non-uniform caliper
of and
rotorcan
andbebrake
largerlining
in comparison
which has toawhen integrated
negative inside
effect on
vehicle handling performance. The air gap between rotor and stator of the most common IWM layout
an IWM.
is shown in Figure 6 as the envelope in which hub bearing deflection is allowed. The gap is also used
to cover
Bearing elastic deformationsfor
manufacturers of this
motor housing
reason due todifferent
develop accelerations, thermomechanical
solutions to reduce theloads,
brake pedal
Sustainability
production 2020,
and12, assembly
4070 4 of 18
travel and improve braketolerances, deformations
caliper piston knock-back due [16].
to press fits and
In case residual
of too largestresses resulting
hub bearing deflection,
from manufacturing
brakes can also endure processes.
non-uniform wear of rotor and brake lining which has a negative effect on
vehicle handling performance. The air gap between rotor and stator of the most common IWM layout
is shown in Figure 6 as the envelope in which hub bearing deflection is allowed. The gap is also used
to cover elastic deformations of motor housing due to accelerations, thermomechanical loads,
production and assembly tolerances, deformations due to press fits and residual stresses resulting
from manufacturing processes.

Figure6.6. Elaphe
Figure ElapheM700
M700IWM with with
IWM centralcentral
hub bearing
hub layout in section
bearing layoutcutinand air gapcut
section in schematic
and air gap in
circumference.
schematic circumference.

TheThe nominalsize
nominal size of
of the
theairairgap
gapforforIWMs
IWMsis normally designed
is normally to be 1tomm.
designed be 1Inmm.
orderIn
toorder
defineto
a define
feasible value for the electromagnetic design, a comprehensive design of the mechanical parts is
a feasible value for the electromagnetic design, a comprehensive design of the mechanical parts is
essential. Ensuring its size is therefore closely connected with electromagnetic performance of the
essential. Ensuring its size is therefore closely connected with electromagnetic performance of the motor
motor and the mechanical torque as an output. One example of air-gap dimensional change in
and the mechanical
correlation thetorque astorque
an output. One example of7,air-gap dimensional change in correlation
Figure 6.with
Elaphe output
M700 IWM withiscentral
presented in Figure
hub bearing layoutwhere the average
in section cut and output
air gap torque is
in schematic
with the output torque is presented in Figure 7, where the average output
calculated in relation to the generated magnetic flux density influenced by the air gap. torque is calculated in
circumference.
relation to the generated magnetic flux density influenced by the air gap.
In addition to these limitations, large diameter seals also require stable operation and consequential
The nominal size of the air gap for IWMs is normally designed to be 1 mm. In order to define a
movement of components within a specific range. For the presented case study, the seal operation field
feasible value for the electromagnetic design, a comprehensive design of the mechanical parts is
is defined to be ±0.2 mm. The presented methods for hub bearing stiffness identification are applicable
essential. Ensuring its size is therefore closely connected with electromagnetic performance of the
and the so-called hub bearing deflection should be known upfront in the design stage in order to
motor and the mechanical torque as an output. One example of air-gap dimensional change in
prevent contact between the static and rotating parts of the motor in the worst-case conditions and to
correlation with the output torque is presented in Figure 7, where the average output torque is
fulfil the requirements for high-diameter sealing. In this study, the objective is directed to automotive
calculated in relation to the generated magnetic flux density influenced by the air gap.
generation
Sustainability 3 hub
2020, bearings
12, x FOR Figureas defined
7. Torque
PEER REVIEW in [17]
output and shown
in relation in size
to air-gap Figure 8. IWM.
for M700 5 of 18

In addition to these limitations, 650 large diameter seals also require stable operation and
Motor torque T [Nm]

600
consequential movement of components 550
within a specific range. For the presented case study, the
seal operation field is defined to be500 ± 0,2 mm. The presented methods for hub bearing stiffness
450
identification are applicable and the so-called hub bearing deflection should be known upfront in the
400
design stage in order to prevent contact 350 between the static and rotating parts of the motor in the
0 0.5
worst-case conditions and to fulfil the requirements for 1 high-diameter
1.5 2 sealing. In this study, the
Air-gap size ag [mm]
objective is directed to automotive generation 3 hub bearings as defined in [17] and shown in Figure
8. Figure7.7.Torque
Figure Torqueoutput
outputininrelation
relationtotoair-gap
air-gapsize
sizefor
forM700
M700IWM.
IWM.

Figure
Figure8. 8.
Hub bearing
Hub bearingsection cutcut
section with schematically
with schematically applied bending
applied moment
bending MxMresulting
moment in
X resulting in
deflection angle β of rotational part in relation to static part.
deflection angle β of rotational part in relation to static part.

Tolerance stack analysis of the complete IWM as shown in Figure 9 should include all
contributing factors to define the maximum allowed deformation due to hub bearing deflection, and
additional research should be made to understand the deflection mechanism to accordingly select or
design a favorable hub bearing.
Sustainability 2020, 12, 4070 5 of 18
Figure 8. Hub bearing section cut with schematically applied bending moment Mx resulting in
deflection angle β of rotational part in relation to static part.
Tolerance stack analysis of the complete IWM as shown in Figure 9 should include all contributing
factors Tolerance stackmaximum
to define the analysis of the complete
allowed IWM as
deformation shown
due in bearing
to hub Figure 9deflection,
should include all
and additional
contributing factors to define the maximum allowed deformation due to hub bearing deflection,
research should be made to understand the deflection mechanism to accordingly select or design a and
additional
favorable hubresearch
bearing.should be made to understand the deflection mechanism to accordingly select or
design a favorable hub bearing.

Figure 9. Radial tolerance stack (TS) path of an IWM with integrated drum brake in section cut. X
Figure 9. Radial tolerance stack (TS) path of an IWM with integrated drum brake in section cut.
markings are presenting press fits without effect on TS.
X markings are presenting press fits without effect on TS.
3. Understandingthe
3. Understanding the Mechanism
Mechanism of
ofBearing
BearingDeflection
Deflection
Double row angular contact ball bearings have a clear advantage regarding axial and radial
Double row angular contact ball bearings have a clear advantage regarding axial and radial
stiffness in comparison to single row ball bearings. They can carry the axial loads in both directions
stiffness in comparison to single row ball bearings. They can carry the axial loads in both directions
whereas offering a substantial level of combined load carrying capacity where the radial, axial and
moment loads act simultaneously. Back-to-back mounting additionally offers high moment rigidity,
which may be further improved by preloading. Due to these advantages, they are used in numerous
applications besides automotive wheel-hubs [18]. Several authors have made investigations and
developed different approaches for performing FEM simulations of bearing deflection in order to
reduce the required time from design to prototyping. All reviews, including results, are noted below in
Table 1.

Table 1. Obtainable research on hub bearing deflection angle done for conventional vehicles, used load
cases and concluding deflections.

Deflection Angle Deflection Angle

Source Applied Load
β-Simulated [◦ ] β-Measured [◦ ]
[19] Fx = 10 kN 0.48 0.49
ax = 0.3 g 0.145 0.161
[20]
ax = 0.6 g 0.229 0.231
[21] Mx = 2 kNm 0.300 0.250
[17] Mx = 3.3 kNm 0.620 0.710
[22] Mx = 1.31 kNm 0.261 0.217

The stiffness concept is derived from the theory of elasticity as the relation between deformation
of the element and input of external loads on the component by:

F = kδ (1)
in Table 1.

Table 1. Obtainable research on hub bearing deflection angle done for conventional vehicles, used
load cases and concluding deflections.

Sustainability 2020, 12, 4070 Deflection Deflection 6 of 18

Source Applied load angle β – angle β –
simulated [°] measured [°]
with F being the applied force, k the stiffness of the component and δ the deformation. This relation
[19] Fx = 10 kN 0.48 0.49
between force and deformation is normally made in a matrix form by:
ax = 0.3 g 0.145 0.161
[20]
ax = 0.6
( g ) "
F1 k 0.229
−k
#(
δ1
) 0.231
[21] Mx = 2 kNm = (2)
F2 −k 0.300
k δ2 0.250
[17] Mx = 3.3 kNm 0.620 0.710
with F1 and F2 being [22]the applied
Mx = 1.31 kNmand δ1 , δ0.261
forces 2 the resulting deformations
0.217 in nodes 1 and 2,
respectively [23]. Stiffness matrix of a double row angular contact ball bearing is obviously more
complex; however,
The stiffness many
concept publications
is derived from theexist that of
theory differ in mathematical
elasticity models
as the relation and
between geometry of
deformation
of analyzed
the elementbearing layouts.
and input Figure 10loads
of external schematically representsby:
on the component the existing approaches for the definition
of the stiffness matrix for a double row ball bearing.
𝐹 = 𝑘𝛿 (1)
The major complexity comes from the nonlinear contact characteristics inside the bearing (between
with
balls andF raceways),
being the applied force, the
which makes k the stiffness
bearing of thenonlinear
stiffness component andand δ the deformation.
load-dependent. In theThis
most
relation
recentbetween forcebearing
publications and deformation
stiffness isisdefined
normallywith made
a5× in5astiffness
matrix form by:(assuming the bearing is
matrix
free to rotate around its axis, so torsional𝐹stiffness 𝑘 is disregarded),
−𝑘 𝛿 for a defined loading condition. For
double row bearings (such as in Figure𝐹10a), = the stiffness may either be identified by two matrices (2)
−𝑘 𝑘 𝛿
(one matrix for each row as shown in Figure 10b), or one stiffness matrix for the double row unit (see
with F1 and F2 being the applied forces and δ1, δ2 the resulting deformations in nodes 1 and 2,
Figure 10c).
respectively [23]. Stiffness matrix of a double row angular contact ball bearing is obviously more
Although the two methods yield identical system deflections, superposing the moment stiffness
complex; however, many publications exist that differ in mathematical models and geometry of
elements of two single rows may not result in the moment stiffness of the double row unit. In this study,
analyzed bearing layouts. Figure 10 schematically represents the existing approaches for the
the latter approach is adopted, as the stiffness of the double row unit dictates the angular displacement.
definition of the stiffness matrix for a double row ball bearing.

Figure 10. (a) Schematic view of an assembled double row bearing [24]; (b) model with two stiffness
Figure 10. (a) Schematic view of an assembled double row bearing [24]; (b) model with two stiffness
matrices; (c) model with one stiffness matrix. (b) and (c) with multi-dimensional non-linear springs
matrices; (c) model with one stiffness matrix. (b) and (c) with multi-dimensional non-linear springs
(no(no
torsional stiffness).
torsional stiffness).

To define the load-deflection characteristic of a single rolling contact, calculation from (1) is
converted into:
Qj = Kn δnj (3)

where Qj represents the resulting normal load on a single rolling element at position j, Kn is a
stiffness constant accounting for geometry and material (also known as Hertzian stiffness constant or
load-deflection factor), and n is a value (exponent) defining the nature of the contact; for point contacts
(i.e., ball bearings) n = 1.5 [18]. By adding the contribution from each rolling element, (3) can be
translated into a complex relationship between the bearing load vector ( fb ) and the bearing deflection
vector (qb ) [25]. Bearing stiffness matrix can then be obtained by applying the mathematical definition
of stiffness and taking partial derivatives of each load term against each deflection term.
 Sustainability 2020, 12, 4070 7 of 18

Deflection (movements and twists) and load (forces and torques) vectors are described by
Equations (4) and (5), respectively, whereas the stiffness matrix is shown by Equation (6), which
included partial translational and rotational (around coordinate axis–elements with θ indexes) stiffness
according to Cartesian coordinate system:
n oT
qb = δx , δ y , δz , βx , β y , (4)

n oT
fb = Fx , F y , Fz , Mx , M y , (5)
 
 kxx kxy kxz kxθx kxθ y 
 

 k yx k yy k yz k yθx k yθ y 

[Kb ]D =  kzx kzy kzz kzθx kzθ y  .

(6)
 
 kθx x kθ y x k θz x kθx θx kθx θ y 
 
kθ y x kθ y y kθ y z k θ y θx kθ y θ y

4. Loads on IWM
Two types of load cases are important for selecting the most suitable bearing for IWM wheel hub
system. The first load case type is the one leading to the most damage in the bearing. It includes
impacts such as driving over an obstacle or a pothole and repetitive loading. The second load case
type is the one leading to the highest bending deflection resulting from severe cornering.
Loads can be prescribed by the application or derived from a test-mule vehicle. In the presented
case study, a BMW X6 was converted with IWM propulsion architecture USING 4X Elaphe L1500
motors with specification shown in performed skid pad test described in ISO 4138-2012 [25].
The vehicle was driving in a circle of radius r = 35.5 m and the speed was increasing with every
lap as shown below in Figure 11. Above v = 70 km/h = 19.44 m/s the vehicle started drifting sideways,
which shows the limit friction of the vehicle. Since this paper is not about tire physics, it can be
assumed that the maximal lateral acceleration is also the maximal tire friction by:
v
µ= = 1.08 (7)
Sustainability 2020, 12, x FOR PEER REVIEW rg 8 of 18

with v being
with thethe
v being vehicle speed,
vehicle r turning
speed, radius
r turning and
radius g gravitational
and acceleration.
g gravitational acceleration.

Figure 11.11.
Figure In-wheel motor
In-wheel vehicle
motor based
vehicle on on
based BMW X6 X6
BMW during skid
during pad
skid testtest
pad at constant radius.
at constant radius.

During
During conversion
conversion of of
thethe
vehicle, thethe
vehicle, ICE,
ICE,gear box,
gear fuel
box, tank,
fuel differential
tank, and
differential transmission
and transmission were
were
removed
removed (−565
(−565kg)kg)
andand4 IWM
4 IWM motors,
motors, 4 inverters
4 inverters andanda battery
a battery pack
packwere
were added
added (+520
(+520kg). This
kg). This
resulted in a modified Centre of Gravity (CoG) position of 131 mm more to the back
resulted in a modified Centre of Gravity (CoG) position of 131 mm more to the back and 72 mm lower and 72 mm lower
than in in
than thetheoriginal
originalstate. Calculated
state. lateral
Calculated forces
lateral on on
forces front tires
front areare
tires presented
presented below
belowin Figure
in Figure12.12.
Forces areare
Forces applied from
applied –2500
from N at N
–2500 7 m/s²
at 7 (0.71
m/s2g) of lateral
(0.71 g) of acceleration on inneron
lateral acceleration front wheel
inner to +13600
front wheel to
N +13,600
at 11 m/s²N (1.12
at 11 g)
m/sof2 lateral
(1.12 g)acceleration on externalonfront
of lateral acceleration wheel
external (Figure
front wheel13). The reduction
(Figure 13). The of CoG
reduction
height decreased
of CoG the maximal
height decreased theload (linked
maximal loadto (linked
maximal toIWM
maximaldeflection) by 7.5%. by 7.5%.
IWM deflection)
removedin(−565
resulted kg) and
a modified 4 IWM
Centre motors,(CoG)
of Gravity 4 inverters andofa131
position battery pack to
mm more were added
the back (+520
and 72 mmkg).lower
This
resulted in a modified Centre of Gravity (CoG) position of 131 mm more to the back and
than in the original state. Calculated lateral forces on front tires are presented below in Figure 12.72 mm lower
than inare
Forces theapplied
original state.
from Calculated
–2500 lateral
N at 7 m/s² (0.71forces on front
g) of lateral tires are presented
acceleration below
on inner front in Figure
wheel 12.
to +13600
Forces are applied from –2500 N at 7 m/s² (0.71 g) of lateral acceleration on inner front wheel
N at 11 m/s² (1.12 g) of lateral acceleration on external front wheel (Figure 13). The reduction of CoG to +13600
N Sustainability
at 11decreased
height m/s² (1.12 g)4070
2020, 12,
the of lateral acceleration
maximal load (linkedon external front
to maximal IWMwheel (Figure
deflection) by13). The reduction of CoG
7.5%. 8 of 18
height decreased the maximal load (linked to maximal IWM deflection) by 7.5%.

Figure 12. Lateral force linked with weight transfer on front wheels depending on the lateral
Figure
Figure12.
12.Lateral
acceleration. Lateral force linkedwith
force linked with weight
weight transfer
transfer on wheels
on front front wheels depending
depending on the
on the lateral lateral
acceleration.
acceleration.

Figure
Figure13.13.Photo
Photofrom one of
from one ofthe
thetest
test sessions,
sessions, where
where accelerations
accelerations were measured
were measured forcornering.
for severe severe
Figure 13. Photo from one of the test sessions, where accelerations were measured for severe
cornering.
cornering.
The loads are within 3% of the maximal loads calculated with multibody dynamics software when
The loads
modelling are within
double 3% of (Figure
lane change the maximal loads
14). This calculated
provides usefulwith multibody
information on dynamics software
the frequency of such
when The loads
modelling are within
double 3%
lane of the
change maximal
(Figure loads
14). Thiscalculated
provides with
usefulmultibody dynamics
information on the software
frequency
extreme events.
Sustainability 2020, 12, x FOR PEER REVIEW 9 of 18
when
of suchmodelling double lane change (Figure 14). This provides useful information on the frequency
extreme events.
of such extreme events.

Figure 14.14.
Figure Simulated version
Simulated forfor
version severe cornering
severe shown
cornering in Figure
shown 12.12.
in Figure

When
Whenloads
loadsare
areapplied
applied in middleofofeach
in the middle eachtire
tire contact
contact patch,
patch, theythey are transferred
are transferred to thetobearing
the
bearing raceways center
raceways center by: by:
M𝑀X =
=𝑀MX−Y++𝑀 MX−Z==𝐹 F𝑅 ++
Y Rw 𝐹 F𝑎Z a (8)(8)
MZ = FX a (9)
𝑀 =𝐹 𝑎 (9)
with MX being the bending moment in longitudinal direction, FY force in the axial direction, Rw
the tire radius, FZ the vertical force on the tire and a, the axial distance between the middle of the
contact patch and the middle of both ball raceways of the bearing, MZ the vertical bending moment
and FX the longitudinal force. The distances Rw and a are not fixed. During cornering the extra weight
 Sustainability 2020, 12, 4070 9 of 18

with MX being the bending moment in longitudinal direction, FY force in the axial direction, Rw the
tire radius, FZ the vertical force on the tire and a, the axial distance between the middle of the contact
patch and the middle of both ball raceways of the bearing, MZ the vertical bending moment and FX the
longitudinal force. The distances RW and a are not fixed. During cornering the extra weight on outer
wheels compress the tires so much that MX-Y decreases (–4% of MX ) and due to the axial deformation
on the tire MX-Z increases (+8% of MX ).
Pothole and vibration-induced fatigue were considered as an unsprung mass system; thus, loading
is replaced by acceleration. For pothole, the complete motor was subjected to two successive 100 g jolts
in both vertical directions. To see if the bearing could sustain vibration, the acceleration-frequency
power spectral density with random signal in accordance to ISO 16750 3:2007 was made as a validation
test upfront [26].

5. Determination of Stiffness
The stiffness of the wheel hub systems is greatly determined by the hub bearing. Its materials
and geometry are the two most important aspects. Obviously, larger bearings have higher stiffness
terms, however not all bearing parameters influence the stiffness in the same way. In fact, the internal
geometry (rolling elements, race ways) rather than basic external geometry (hub bearing housing)
affects the bearing stiffness terms. For ball bearings, these parameters are the pitch diameter, ball
diameter, number of balls in each row, contact angle, and radii of the inner and outer raceway curvatures
(raceway conformities). Thus, it is important to know the internal design details of the bearing to
accurately estimate the bearing stiffness.
Bearing mounting arrangement and row separation distance D (Figure 15) are two other important
design parameters, as these parameters determine the effective spread E (Figure 15) together with
the pitch diameter and contact angle. The effective spread is a measure of the moment stiffness of
the bearing. Back-to-back or O-arrangement bearings have much larger spread than face-to-face
X-arrangement bearings (Figure 15). Due to this reason, automotive hub bearings are always mounted
back-to-back
Sustainability (O).
2020, 12, x FOR PEER REVIEW 10 of 18

Figure 15. 15.

Figure Illustration of back-to-back
Illustration (O)(O)
of back-to-back andand
face-to-face (X) (X)
face-to-face bearing arrangement.
bearing arrangement.

The moment
In order stiffness
to evaluate the of the hubeffects
resulting bearingofmay be further
all these improved
parameters by preloading.
combined, This isbearing
an analytical discussed
in more detail in the following section.
design/analysis tool EBECA (Elaphe BEaring Calculator) has been developed (graphical interface
shown Inin order
Figureto16).
evaluate the resulting
This software effectsfor
prompts of the
all these parameters
bearing geometrycombined, an analytical
and mounting bearing
information
design/analysis tool EBECA (Elaphe BEaring Calculator) has been developed
(mentioned above), along with the bearing preload and load vectors acting on the bearing. Due to(graphical interface
shown in Figure
nonlinearity, 16). This
an iterative methodsoftware
based prompts for thealgorithm
on the simplex bearing geometry andtomounting
is then used output theinformation
bearing
(mentioned
deflection above),
vector, along
stiffness withand
matrix the internal
bearing load
preload
andand load
stress vectors acting on the bearing. Due to
distributions.
nonlinearity, an iterative method based on the simplex algorithm is then used to output the bearing
deflection vector, stiffness matrix and internal load and stress distributions.
 Sustainability 2020, 12, 4070 10 of 18

Figure 15. Illustration of back-to-back (O) and face-to-face (X) bearing arrangement.

In this example, the data of the BMW X6 front hub bearing were fed into the interface. Repeating
In order to evaluate the resulting effects of all these parameters combined, an analytical bearing
this analysis for several operating points and for various bearings, the deflection results are obtained
design/analysis tool EBECA (Elaphe BEaring Calculator) has been developed (graphical interface
as shown in Figure 17. For benchmark purposes, several other conventional hub bearings have also
shown in Figure 16). This software prompts for the bearing geometry and mounting information
been analyzed.
(mentioned above), along with the bearing preload and load vectors acting on the bearing. Due to
EBECA
nonlinearity, analytical
an iterative bearing
method toolon
based is the
used to determine
simplex load
algorithm distribution
is then and stresses
used to output within the
the bearing
bearingvector,
deflection shownstiffness
in Figure 18. and internal load and stress distributions.
matrix

Figure 16. Input

Sustainability 2020, data for PEER
12, x FOR standard BMW X6 front bearing in the analytical bearing design/analysis
REVIEW 11 of 18
Figure 16. Input data for standard BMW X6 front bearing in the analytical bearing design/analysis tool.
tool.

In this example, the data of the BMW X6 front hub bearing were fed into the interface. Repeating
this analysis for several operating points and for various bearings, the deflection results are obtained
as shown in Figure 17. For benchmark purposes, several other conventional hub bearings have also
been analyzed.

Figure
Figure 17.17.Analytically
Analytically calculated
calculated bearing
bearingdeflection without
deflection including
without the elastic
including deformation
the elastic of
deformation of
bearing housing.
bearing housing.

EBECA analytical bearing tool is used to determine load distribution and stresses within the
bearing shown in Figure 18.
 Sustainability 2020, 12, 4070 11 of 18

To take into account static and dynamic flanges deformations the bearing geometry and load were
Figuresimulated
17. Analytically calculated
using Finite Elementbearing
Methodsdeflection
(FEM). Thewithout including
deformation shownthe elastic
is not to be deformation
neglected, as itof
bearing housing.
accounts for an overall contribution of 30% of the total angular deflection, as seen below in Figure 19.
The analysis also confirms that the highest stresses near the bolt threads are below the element’s Yield
EBECA analytical
point. bearing tool
Elastic deformation of hubisbearing
used housing
to determine
elementsload distribution
has been numerically and stresses
simulated within
for several the
load cases and added
bearing shown in Figure 18. to the calculated bearing deflections for several hub bearings in the later sections.

Figure
Sustainability 2020,18. xLoad
Figure
12, distribution
FOR18. Load
PEER calculated
distribution
REVIEW using
calculated usingbearing design/analysis
bearing design/analysis tool.tool. 12 of 18

To take into account static and dynamic flanges deformations the bearing geometry and load
were simulated using Finite Element Methods (FEM). The deformation shown is not to be neglected,
as it accounts for an overall contribution of 30% of the total angular deflection, as seen below in Figure
19. The analysis also confirms that the highest stresses near the bolt threads are below the element’s
Yield point. Elastic deformation of hub bearing housing elements has been numerically simulated for
several load cases and added to the calculated bearing deflections for several hub bearings in the later
sections.

Figure19.
Figure 19. Visualization
Visualization of
of results
results for
for aa numerical
numerical simulation
simulation of
of BMW
BMW X6X6 M
M front
front bearing
bearing Y-plane
Y-plane
deformationin
deformation in[mm]
[mm]with
withmaximum
maximumbending
bendingmoment
momentapplied
appliedaround
aroundYYaxis.
axis.

6.6. Validation
Validationof
ofHub
HubBearing
BearingStiffness
Stiffnesson
onTest
TestRig
Rig
In
In order
order to
to validate
validate the
the bearing
bearing stiffness
stiffness calculations,
calculations, several
several bearing
bearing deflection
deflection tests
tests were
were
performed
performed on a custom-made bearing deflection test rig (Figure 20). Bearing stiffness propertieswere
on a custom-made bearing deflection test rig (Figure 20). Bearing stiffness properties were
derived
derivedfrom
fromthe
thebearing
bearingrotor
rotorflange
flangedisplacements
displacementsmeasured
measuredwithwithlaser
lasertriangulation
triangulationand
andthetheload
load
applied with hydraulic cylinders.
applied with hydraulic cylinders.

Figure 20. BMW X6 front bearing on the deflection test rig at the University of Ljubljana (UL), Faculty
 deformation in [mm] with maximum bending moment applied around Y axis.
6. Validation of Hub Bearing Stiffness on Test Rig
6. Validation of Hub Bearing Stiffness on Test Rig
In order to validate the bearing stiffness calculations, several bearing deflection tests were
In order
performed on to validate the bearing
a custom-made bearingdeflection
stiffness calculations, several
test rig (Figure bearing
20). Bearing deflection
stiffness tests were
properties were
performed on
Sustainability a custom-made
2020, 12, 4070 bearing deflection test rig (Figure 20). Bearing stiffness
derived from the bearing rotor flange displacements measured with laser triangulation and the 12 properties were
load
of 18
derived from the bearing rotor
applied with hydraulic cylinders. flange displacements measured with laser triangulation and the load
applied with hydraulic cylinders.

Figure 20.20.
Figure BMWBMWX6X6
front bearing
front onon
bearing thethe
deflection test
deflection rigrig
test at at
thethe
University ofof
University Ljubljana (UL),
Ljubljana Faculty
(UL), Faculty
Figure
ofof 20. BMW
Mechanical X6 front bearing
Engineering (FME) on the deflection test rig at the University of Ljubljana (UL), Faculty
[27].
Mechanical Engineering (FME) [27].
of Mechanical Engineering (FME) [27].
Tests
Testswere
wereperformed
performedfor
forthree
threedifferent
differentload
loadcases
casesacting
actingononthe
thehub
hubbearing,
bearing,asasdescribed
describedinin
Tests
Figure
Figure21.
21.were performed for three different load cases acting on the hub bearing, as described in
Figure 21.

Figure 21. Applied load cases (LC) for performing deflection tests at FME.
Figure 21.21.
Figure Applied load
Applied cases
load (LC)
cases forfor
(LC) performing deflection
performing tests
deflection at at
tests FME.
FME.
For bearing performance, particularly in the range of bearing stiffness, the bearing preload plays
For
a major bearing
For
role [28].performance,
bearing performance,
This particularly
preload onparticularlyininrolling
the bearing the
therange
rangeofofbearing
elementsbearing stiffness,
stiffness,
is typically the
thebearing
bearing
achieved preload
preload
by closing plays
theplays
gap a
aormajor
major role [28]. This
play between This preload
the preload on the bearing rolling
two raceway rings. Depending elements
elements is typically
on theis bearing achieved
typicallydesign,
achieved by closing the
by closing the
gap may gap
gapbeor
orplay
playbetween
between thethe two
two raceway
raceway rings.
rings. Depending
Depending onbearing
on the the bearing
design,design, gap closing
gap closing may bemay be
obtained
during the manufacturing process or applied during the bearing installation. Sustaining a sufficient
bearing preload under all working conditions is critical and, in this case, ensured with sufficient axial
prestress. This is ensured with a sufficient level of external clamping force. Influence of the amount
of clamping force was evaluated with the variation of bearing tightening torque. This parameter is
crucial when designing the IWM preload shaft used for generating the clamping force.
Bearing deflection measurements for the maximum bending moment are presented in Figure 22.
The load was limited to 4500 Nm (except for Smart For Two bearing with low load capacity), which is
generated by three different ways representing the three load cases:
Load Case 1 with the use of only one hydraulic cylinder generating Fk2 , resulting in 7.2 kN of
compressive axial force and 4500 Nm of bending moment acting on the bearing.
Load Case 2 with the use of both hydraulic cylinders generating Fk1 and Fk2 , each applying a
force up to 3.6 kN resulting in 4500 Nm of pure bending moment acting on the bearing.
Load Case 3 with the use of only one hydraulic cylinder generating Fk1 , resulting in 7.2 kN
of tensile axial force and 4500 Nm of bending moment acting on the bearing. Figure 22 shows
measurements for load case 2, which results in the highest deformations from all three scenarios.
 Load Case 1 with the use of only one hydraulic cylinder generating Fk2, resulting in 7.2 kN of
compressive axial force and 4500 Nm of bending moment acting on the bearing.
Load Case 2 with the use of both hydraulic cylinders generating Fk1 and Fk2, each applying a
force up to 3.6 kN resulting in 4500 Nm of pure bending moment acting on the bearing.
Load Case 3 with the use of only one hydraulic cylinder generating Fk1, resulting in 7.2 kN of
tensile axial
Sustainability 2020, force
12, 4070and 4500 Nm of bending moment acting on the bearing. Figure 22 shows 13 of 18
measurements for load case 2, which results in the highest deformations from all three scenarios.

Figure
Sustainability
Figure 22.Bearing
2020,
22. Bearing
12, x FORdeflection measurement
PEER REVIEW
deflection measurement results for for
results different hub hub
different bearings, original
bearings, in specified
original 14 of 18
in specified
vehicles and pure bending load case [1].
vehicles and pure bending load case [1].
counter moment of force is of the order of magnitude of 1% of the bending moment resulting from
7.
theValidation ofofthe
road; thus,
7. Validation IWM
IWM Stiffness
stiffness contribution from magnets is small and can be neglected.
Stiffness
The
In thetest
1980s,procedure
the need arosewas fordefined as wheel
testing the follows: first,components
related the IWM instiffness was measured
a fast, repetitive and safe
In the 1980s, the
(characterization need arose thenfor testing the wheel related components in a fast,defined
repetitive and safe
way. Mechanical sequence),
testing systems the
werehighest
built,positive and
with which negative bendingdamage
the cumulative moments in the
over the whole
way. Mechanical
requirements testing systems were built, with which the cumulative damage over the whole lifetime
lifetime couldwere tested five
be reliably times, alternating
reproduced within a in thehours
few cornering
[29].sequence. The third
These machines sequence
were tested
designed at
could be reliably
bearing durability
Fraunhofer reproduced
Institute andforwas within a few
based onDurability
Structural hours [29].
the SAE international These
and System J328/2005-2 machines
Reliability LBF were
standard designed
[30]. The
for testing load
tires, at Fraunhofer
onand
rims the
Institute
bearing for
wasStructural
increased Durability
by 250% and
and System
the hence Reliability
duration LBF
of for
the testing
acceleratedtires,
bearings. IWM is a novel component that can greatly benefit from such intensive testing. In our case rims
lifetime and
test bearings.
could be IWM
reduced
is athe
novel to 26 hours. The characterization and cornering sequences were repeated
Wheel Accelerated Life Test machine (W/ALT) from Fraunhofer LBF is used to verify bearing Wheel
component that can greatly benefit from such intensive testing. after
In our the fatigue
case the
tests each time.
Accelerated
durability, Life
housingTestelastic
machine (W/ALT)
deformation andfrom Fraunhofer
the air-gap LBF isduring
deformation used operation
to verify(Figure
bearing durability,
23). It
During
generates the entire
a realistic procedure,
wheel contact thermal
and sensors
side force were placed
conditions using aonsix-axis
the bearing hub and cylinders.
set of(Figure
hydraulic a thermal
housing elastic deformation and the air-gap deformation during operation 23). It generates a
camera filmed the test. Moreover,forces
two laser sensors measured the axial and radial deformation of the
realisticThe
wheelradial electromagnetic
contact in the
and side force conditions IWM resist
using a six-axisdeflection bending,
set of hydraulic which can
cylinders. be
rotor.
described with an additional moment of force. This counter moment MX-Mag can be integrated over all
magnets, presuming a pure bending of the rotor by:

𝑀 (𝛽) = 𝐹 𝑍 sin(𝛼) (13)

with β being the angular deflection, N is the number of magnets, FYi is the axial force of i-th
magnet, and Zi is the vertical position of i-th magnet. After integration the results show that this
counter moment of force is of the order of magnitude of 1% of the bending moment resulting from
the road; thus, the stiffness contribution from magnets is small and can be neglected.

Figure 23. IWM in Wheel Accelerated Life test (WALT) for endurance testing at Fraunhofer LBF [1].
Figure 23. IWM in Wheel Accelerated Life test (WALT) for endurance testing at Fraunhofer LBF [1].

The stiffness of the motor is derived from rotor displacement measurements based on the
The radial electromagnetic forces in the IWM resist the deflection bending, which can be described
applied loads. The resulting stiffness is not constant but increases with the bending moment. When
with an additional moment of force. This counter moment MX-Mag can be integrated over all magnets,
extremely loaded, the bending stiffness of the bearing can be twice as high as when not loaded. This
presuming a pure
is due to the bending of the
load-dependency rotor
of the by: matrix elements as discussed earlier.
stiffness
The angular deflection of the rotor is plotted in Figure 24. It can be seen that the difference
=N
iX
between before and after the endurance sequence is within 8%.
MX−Mag (β) = FYi Zi sin(α) (10)
i=1

with β being the angular deflection, N is the number of magnets, FYi is the axial force of i-th magnet,
and Zi is the vertical position of i-th magnet. After integration the results show that this counter
moment of force is of the order of magnitude of 1% of the bending moment resulting from the road;
thus, the stiffness contribution from magnets is small and can be neglected.
The test procedure was defined as follows: first, the IWM stiffness was measured (characterization
sequence), then the highest positive and negative bending moments defined in the requirements were
tested five times, alternating in the cornering sequence. The third sequence tested bearing durability
and was based on the SAE international J328/2005-2 standard [30]. The load on the bearing was
 (characterization sequence), then the highest positive and negative bending moments defined in the
requirements were tested five times, alternating in the cornering sequence. The third sequence tested
bearing durability and was based on the SAE international J328/2005-2 standard [30]. The load on the
bearing was increased by 250% and the hence duration of the accelerated lifetime test could be
reduced to 26 hours. The characterization and cornering sequences were repeated after the fatigue
Sustainability
tests each2020, 12, 4070
time. 14 of 18
During the entire procedure, thermal sensors were placed on the bearing hub and a thermal
camera filmed the test. Moreover, two laser sensors measured axial and radial deformation of the
increased
rotor. by 250% and the hence duration of the accelerated lifetime test could be reduced to 26 h.
The characterization and cornering sequences were repeated after the fatigue tests each time.
During the entire procedure, thermal sensors were placed on the bearing hub and a thermal
camera filmed the test. Moreover, two laser sensors measured axial and radial deformation of the rotor.
The stiffness of the motor is derived from rotor displacement measurements based on the applied
loads. The resulting stiffness is not constant but increases with the bending moment. When extremely
loaded, the bending stiffness of the bearing can be twice as high as when not loaded. This is due to the
load-dependency of the stiffness matrix elements as discussed earlier.
The angular deflection of the rotor is plotted in Figure 24. It can be seen that the difference between
before and after the endurance sequence is within 8%.
During the23.test
Figure IWMnoincomponents wereLife
Wheel Accelerated damaged, neither
test (WALT) did a contact
for endurance testing atin the air gap
Fraunhofer LBF occur.
[1]. Thermal
sensors showed that the bearing never started to overheat. The repetition of the five highest bending
momentsThe stiffness of the motor is derived from rotor displacement measurements based on the
is visible on the temperature measurements of the bearing (see Figure 25). During the long
applied loads. The resulting stiffness is not constant but increases with the bending moment. When
endurance sequence test, the temperature is raised until equilibrium, which is reached in 15 min.
extremely loaded, the bending stiffness of the bearing can be twice as high as when not loaded. This
The is
motor’s temperature
due to the is shown
load-dependency at stiffness
of the 0, 5 andmatrix
15 min in Figure
elements 26. Afterwards,
as discussed earlier. the temperature of the
hub bearing remained
The angular constant,
deflection which
of the rotorindicates
is plottedthat no damage
in Figure 24. It has occurred
can be that
seen that thecould increase the
difference
residual torque.
between before and after the endurance sequence is within 8%.

Sustainability 2020, 12, x FOR PEER REVIEW 15 of 18

the long endurance sequence test, the temperature is raised until equilibrium, which is reached in 15
minutes. The motor’s temperature is shown at 0, 5 and 15 minutes in Figure 26. Afterwards, the
temperature of the hub bearing
Measurements
Figure 24. 24. of remained
the constant,
deflection which indicates
for analyzed IWM with that no damage
BMW X6 hubhas occurred
bearing thatand
before
couldFigure
increase Measurements of the deflection for analyzed IWM with BMW X6 hub bearing before and
the residual torque.
afterafter
the the
endurance
endurancesequence
sequence[1].
[1].

During the test no components were damaged, neither did a contact in the air gap occur.
Thermal sensors showed that the bearing never started to overheat. The repetition of the five highest
bending moments is visible on the temperature measurements of the bearing (see Figure 25). During

Figure
Figure 25. 25. Temperaturemeasurements
Temperature measurements with withtimestep 1 ms
timestep on the
1 ms on hub
the bearing, directlydirectly
hub bearing, above ball
above ball
(green)
(green) andand 15 mm
15 mm away
away (blue)during
(blue) during the
the first
first characterization
characterization and cornering
and sequences
cornering as shown
sequences as shown on
on Figure 8.
Figure 8.
 Figure 25. Temperature measurements with timestep 1 ms on the hub bearing, directly above ball
(green) and
Sustainability 1512,
2020, mm away (blue) during the first characterization and cornering sequences as shown 15 of 18
4070
on Figure 8.

Figure 26. 26.

Figure Thermal pictures
Thermal of the
pictures of motor whenwhen
the motor the radial endurance
the radial sequence
endurance starts starts
sequence (from (from
0 to 150 to
min) [1].
15 min) [1].

8. Results
8. Results
TheThe performed
performed tests
tests successfully
successfully demonstrate
demonstrate that
that thethe bearing
bearing cancan sustain
sustain thethe
loadload
andand prevent
prevent
contact in the air gap, which was the main objective. Additionally, bearing deflection tests before andand
contact in the air gap, which was the main objective. Additionally, bearing deflection tests before
after
after fatigue
fatigue tests
tests showshow that
that fatigue
fatigue tests
tests diddid
notnot result
result in hub
in hub bearing
bearing damage,
damage, which
which is also
is also vital
vital forfor
thethe definition
definition of of
the the productlifetime
product lifetimeand
andmaintenance
maintenanceinterval.
interval. Figure
Figure 27 shows high
high correlation
correlationofofthe
thedeflection
deflectionteststestsmade
made onon
IWM,
IWM, hubhub
bearing andand
bearing the predicted characteristics
the predicted of theofanalytical
characteristics bearing
the analytical
design/analysis
bearing design/analysistool. tool.
Sustainability 2020, 12, x FOR PEER REVIEW 16 of 18

Figure 27.27.
Figure BMW X6 X6
BMW bearing deflection
bearing comparison.
deflection comparison.

The
Thehigher calculated
higher bearing
calculated bearing stiffness compared
stiffness compared to to
thethe
measured
measured one cancan
one bebeexplained
explainedbyby
thethe
elasticity of of
elasticity hub bearing
hub housing
bearing housingcomponents
components and also
and IWM
also IWM stator and
stator rotor
and plates,
rotor bolt
plates, connections
bolt connections
as as
well as as
well thethe
adapters used
adapters forfor
used fixating
fixatingthethe
bearing
bearingduring
during thethe
deflection tests.
deflection tests.

9. Conclusions
9. Conclusion
From
From thethe design
design perspective,
perspective, the the approach
approach presented
presented herehere drastically
drastically simplifies
simplifies the selection
the selection of
of a suitable hub bearing geometry for a specific in-wheel motor application
a suitable hub bearing geometry for a specific in-wheel motor application with the developed with the developed
analytical
analytical bearing
bearing design/analysis
design/analysis tool.
tool. TheThe
hubhub bearing,
bearing, being
being a critical
a critical component
component forfor in-wheel
in-wheel
motors,
motors, hasmore
has moredemanding
demanding requirements
requirements than
than wheels
wheelsininconventional
conventional propulsion architectures,
propulsion since
architectures,
the deflection angle affects the motor functionality and output torque. The proposed
since the deflection angle affects the motor functionality and output torque. The proposed approach approach and
anddeveloped
developedinternal EBECA
internal EBECAtool tool
allowallow
engineers to design
engineers a functional
to design IWM assembly
a functional faster, predict
IWM assembly faster,the
predict the operating turning moment (torque) envelope within a specified application related to
driving regime, and defines testing procedure with fatigue lifetime tests. In addition to the hub
bearing importance, it must be mentioned at this point that the housing design also has a significant
impact on the IWM stiffness, as comprehensively covered in [31].
Sustainability 2020, 12, 4070 16 of 18

operating turning moment (torque) envelope within a specified application related to driving regime,
and defines testing procedure with fatigue lifetime tests. In addition to the hub bearing importance,
it must be mentioned at this point that the housing design also has a significant impact on the IWM
stiffness, as comprehensively covered in [31].

10. Patents
1. WO2012138303A2; Electromagnetic design: Compact multiphase wave winding of a high specific
torque electric machine.
2. WO2018124971A1; 23465; Electromagentic design of in-wheel motors: Arrangement for
determining maximum allowable torque.
3. CT/EP2017/081085; WO/2012/138303/A2; Electric machine with a cooling system and a method
for cooling an electric machine.
4. SI23465; WO/2013/180663; Electrical gear for electric vehicles with direct drive.
5. SI23406; Electric machine with reduced holding torque, with torque vibration and unchanged
torque constant.
6. PCT/EP2017/079793; WO/2018/095868; Integrated electric gear and charger system for battery
powered electric vehicles.
7. EP3340439; USA: 20180183292 and EPO: 3340439; Voltage balanced winding pattern for an electric
machine with a minimal number of connections and method for assembly of such winding.
8. PCT/SI2016/000030; WO/2018/124971; Arrangement for determining maximum allowable torque.
9. WO/2019/098949; Method and apparatus for compact insertion of multiphase pseudo helical
wave winding into electrical machine.
10. WO/2019/139545; In-wheel electric motor maintenance integration.
11. WO/2019/151956; Integrated gap retention element for electric motor.
12. WO/2020/00966; Electric vehicle energy balance crediting and debiting system and a
method thereof.

Author Contributions: M.B. coordinated the activity, in charge of research background, definition of required
tests, correlated activities on electromagnetic responses and functionality from the durability point of view. R.C.
identification of acting loads on the in-wheel motor component and definition of load cases and execution of
validation test at Fraunhofer LBF. A.G. development and validation of stiffness matrix formulation for double row
angular contact ball bearings. S.O. development of bearing stiffness calculator and calculation of stiffness matrices.
Definition of Definition and validation tests. S.Z. PI and experimental tests and validation of hub bearing stiffness
matrix. R.K. experimental measurements and validation of hub bearing stiffness matrix. All authors have read
and agreed to the published version of the manuscript.
Funding: The work presented in this article was partially funded by the Slovenian Research Agency as part of the
“Modelling in technics and medicine” (code P2-0109) research programme.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Elaphe Propulsion Technologies, Ltd. Internal archive (formats: .pdf, .cae, .odb, .sldprt, .sldasm, .stp, .sat); Elaphe
Propulsion Technologies, Ltd.: Ljubljana, Slovenia, 2017.
2. Vallance, A. Advanced In-Wheel Electric Propulsion Technology. Protean Electric, The Centre For Sustainable
Design. 2010. Available online: https://cfsd.org.uk/pdf/March%202010%20General%20Presentation%20-
%20CFSD%20lecture%204.pdf (accessed on 15 June 2019).
3. Perovic, D.K. Making the Impossible, Possible–Overcoming the Design Challenges of in Wheel Motors.
World Electr. Veh. J. 2016, 5, 514–519. [CrossRef]
4. Ifedi, C.J.; Mecrow, B.C.; Brockway, S.T.M.; Boast, G.S.; Atkinson, G.J.; Kostic-Perovic, D. Fault tolerant
in-wheel motor topologies for high performance electric vehicles. In Proceedings of the 2011 IEEE International
Electric Machines Drives Conference, Niagara Falls, ON, Canada, 14–17 May 2010; pp. 1310–1315.
Sustainability 2020, 12, 4070 17 of 18

5. Cakir, K. In-Wheel Motor Design for Electric Vehicles. Ph.D. Thesis, Sabanci Unversity, Tuzla, Turkey, 2004.
6. Gruber, W.; Back, W.; Amrhein, W.; Bäck, W. Design and implementation of a wheel hub motor for an electric
scooter. In Proceedings of the 2011 IEEE Vehicle Power and Propulsion Conference, Chicago, IL, USA, 6–9
September 2011; pp. 1–6.
7. Pérez, S.R. Analysis of a Light Permanent Magnet In-Wheel Motor for an Electric Vehicle with Autonomous Corner
Modules; Royal Institute of Technology (KTH): Stockholm, Sweden, 2011.
8. Heim, R.; Hanselka, V.; El Dsoki, C. Technical potential of in-wheel motors. ATZ 2012, 114, 4–9.
9. Wang, R.; Wang, J. Fault-Tolerant Control with Active Fault Diagnosis for Four-Wheel Independently Driven
Electric Ground Vehicles. IEEE Trans. Veh. Technol. 2011, 60, 4276–4287. [CrossRef]
10. Vos, R. Influence of In-Wheel Motors on the Ride Comfort of Electric Vehicles; Eindhoven University of Technology:
Eindhoven, The Netherlands, 2010.
11. Fraser, A. In-Wheel Electric Motors. In Proceedings of the 10th International CTI Symposium, Novi, MI,
USA, 9–12 May 2016.
12. Nagaya, G. Development of an in-wheel drive with advanced dynamic-damper mechanism. JSAE Rev. 2003,
24, 477–481. [CrossRef]
13. Biček, M.; Gotovac, G.; Miljavec, D.; Zupan, S. Mechanical Failure Mode Causes of In-Wheel Motors. Strojniški
Vestnik J. Mech. Eng. 2015, 61, 74–85. [CrossRef]
14. Ziehl-Abegg Automotive GmbH. ZAwheel—Movement by perfection, ZAwheel In-Wheel Hub Motor. 2014.
Available online: http://www.ziehl-abegg.com/gb/en/product-range/automotive/ (accessed on 30 July 2019).
15. Wong, J.Y. Theory of Ground Vehicle, 3rd ed.; John Wiley & SONS, Inc.: New York, NY, USA, 2001.
16. Shevket, C. Asymmetric Hub Assembly. U.S. Patent 10964013, 13 October 2004.
17. NTN. Hub Bearings—CAT. No. 4601/E, CAT. No. 4601/E. 2014. Available online: http://www.
ntnamericas.com/en/website/documents/brochures-and-literature/catalogs/hub_bearings_4601.pdf (accessed
on 15 June 2019).
18. Harris, T.A.; Kotzalas, M.N. Essential Concepts of Bearing Technology, 5th ed.; Taylor & Francis: Boca Raton, FL,
USA, 2007.
19. Yang, M. A Study on the Lateral Stiffness of the Passenger Car Suspension. In Proceedings of the 2012
SIMULIA Customer Conference, Providence, RI, USA, 15–17 May 2012; pp. 1–21.
20. Koyama, T. Applying FEM to the Design of Automotive Bearings. Motion Control 1997, 2, 23–30.
21. Kajihara, K. Improvement of Simulation Technology for Analysis of Hub Unit Bearing. KOYO Eng. J. Engl.
Ed. 2005, 167, 35–39.
22. Lee, I.; Cho, Y.; Cho, Y.; Kim, M.; Jang, C.; Lee, Y.; Lee, S. Development of Stiffness Analysis Program for
Automotive Wheel Bearing. In Proceedings of the 2012 SIMULIA Customer Conference, Providence, RI,
USA, 15–17 May 2012.
23. Rothbart, H.; Brown, T.H. Mechanical Design Handbook, 2nd ed.; McGraw-Hill: New York, NY, USA, 2006.
24. Gunduz, T.H.; Singh, R. Stiffness matrix formulation for double row angular contact ball bearings: Analytical
development and validation. J. Sound Vib. 2013, 332, 5898–5916. [CrossRef]
25. ISO. ISO 4138:2012—Passenger Cars—Steady-State Circular Driving Behavior—Open-Loop Test Methods; ISO:
Geneva, Switzerland, 2012.
26. ISO. ISO 16750-3:2007(E)—Road Vehicles—Environmental Conditions and Testing for Electrical and Electronic
Equipment—Part 3: Mechanical Loads; ISO: Geneva, Switzerland, 2007.
27. Frajnkovic, M.; Omerovic, S.; Rozic, U.; Kern, J.; Connes, R.; Rener, K.; Biček, M. Structural Integrity of
In-Wheel Motors. In Proceedings of the International Powertrains, Fuels & Lubricants Meeting, Heidelberg,
Germany, 18 September 2018; Volume 1, p. 11.
28. Gunduz, A.; Dreyer, J.T.; Singh, R. Effect of bearing preloads on the modal characteristics of a shaft-bearing
assembly: Experiments on double row angular contact ball bearings. Mech. Syst. Signal Process. 2012, 31,
176–195. [CrossRef]
29. Gerhard, F.; Grubisić, V.; Fischer, G.; Grubisić, V.; Grubišić, V. Biaxial Wheel/Hub Test Facility; Report No.
TB-221; Fraunhofer Institute for Structural Durability and System Reliability LBF Darmstadt: Darmstadt,
Germany, 2001.
Sustainability 2020, 12, 4070 18 of 18

30. SAE International. SAE International J328/2005-2—Wheels; Passenger Car and Light Truck Performance
Requirements and Test Procedures; SAE International: Warrendale, PA, USA, 2009.
31. Biček, M. Modelling and Optimization of In-Wheel Motor Mechanical Design. Ph.D. Thesis, University of
Ljubljana, Ljubljana, Slovenia, 2019.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).