Wear 311 (2014) 7180

Contents lists available at ScienceDirect

Wear
journal homepage: www.elsevier.com/locate/wear

A comprehensive model of wear, friction and contact temperature

in radial shaft seals$
D. Frlich n, B. Magyar, B. Sauer
Institute of Machine Elements, Gears, and Transmissions (MEGT), University of Kaiserslautern-Gottlieb-Daimler-Strae 42, 67663 Kaiserslautern, Germany

art ic l e i nf o

a b s t r a c t

Article history:
Received 3 September 2013
Received in revised form
26 December 2013
Accepted 31 December 2013
Available online 9 January 2014

Radial shaft seals are used in a variety of applications, where rotating shafts in steady housings have to be
sealed. Typical examples are crankshafts, camshafts, differential gear or hydraulic pumps. In the
operating state the elastomeric seal ring and the shaft are separated by a lubrication lm of just a few
micrometers. Due to shear strain and uid friction the contact area is subject to a higher temperature
than the rest of the seal ring. The stiffness of the elastomeric material is intensely inuenced by this
temperature and thus contact pressure, friction and wear also strongly depend on the contact
temperature. In order to simulate the contact behavior of elastomer seal rings it is essential to use a
comprehensive approach which takes into consideration the interaction of temperature, friction and
wear. Based on this idea a macroscopic simulation model has been developed at the MEGT. It combines a
nite element approach for the simulation of contact pressure at different wear states, a semi-analytical
approach for the calculation of contact temperature and an empirical approach for the calculation of
friction. In this paper the model setup is presented, as well as simulation and experimental results.
& 2014 Elsevier B.V. All rights reserved.

Keywords:
Radial shaft seal ring
Contact temperature
Wear
Friction torque
Finite element simulation

1. Introduction
The eld of applications for elastomeric radial shaft seals is
wide and diverse. Automotive industry, washing machines, water
pumps or wind power plants are just four examples that emphasize how fundamentally different the applications and thus the
requirements for seals can be. A typical radial shaft seal design is
illustrated in Fig. 1.
In some applications, temperature is the critical factor. Especially where large diameters and circumferential speeds are
involved, the thermal properties of the elastomer need to be taken
into consideration. In other applications, the degree of efciency
and the reduction of friction are signicant. In all elds of use, the
seal rings need to outlast the lifetime of the machine they are
installed in. Thus ageing and wear must be reduced to a minimum.
Due to this variety of applications and requirements, the choice of
an appropriate seal ring design and material is a challenge.
During the development process of new seal designs or new
elastomer materials, test rigs are generally used to check the seal
ring under conditions that are close to the designated eld of use.
Experiments focusing on wear are especially time consuming. This
is the reason why in recent years, simulation models have been
increasingly used for preliminary tests of the seal ring behavior
and to reduce the experimental expense.
$
n

This paper was presented at the 2013 World Tribology Congress.

Corresponding author. Tel.: 49 631 205 3715.
E-mail address: daniel.froelich@mv.uni-kl.de (D. Frlich).

0043-1648/$ - see front matter & 2014 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.wear.2013.12.030

In a rst section, this paper will give an overview of different

simulation models for radial shaft seal rings that have been
developed in companies or research institutes in recent years.
Then the model that has been developed at the Institute of
Machine Elements, Gears and Transmissions (MEGT) will be
explained. The structure of the model is introduced and a
comparison with experimental results is given.

2. State of the art

Many researchers have developed simulation models to investigate the behavior of seal rings in detail. Basically it can be
distinguished between macroscale and microscale modelling
approaches. The models making use of microscale approaches
typically focus on a small section of the seal system that is
modelled physically detailed and very close to reality. These
models can be used to extend the fundamental understanding of
the functionality of radial shaft seal rings. Macroscale models are
often based on simplied empirical approaches, combined with
higher scale physical models. The focus of these macroscale
models is on the function of the overall system.
Below the state of the art of these seal ring models is presented.
Most of the macroscale models focus on just one aspect, that is,
on temperature, friction, or wear. By the use of an electro-thermal
analogy-model Upper [1] is able to determine the contact temperature in a seal ring as well as the temperature distribution.
It can be shown that the wear of a seal ring and the associated

72

D. Frlich et al. / Wear 311 (2014) 7180

Fig. 1. Schematic of radial shaft seal ring components.

growing contact width have a signicant inuence on the temperature distribution within the seal ring. A high temperature
gradient within the seal ring volume is detected. Based on these
scientic ndings, rst nite element models that combine thermal and mechanical load steps are developed.
Stakenborg and Ostayen investigate the temperature-distribution in the contact zone of radial shaft seal rings [2] with the
nite element method and a thermal network. The authors demonstrate that the larger proportion of the heat generated in the contact
area dissipates through the shaft and a considerably smaller proportion dissipates through the seal ring.
Kang and Sadeghi use a microscale model to simulate the
temperature-distribution in the contact zone of seal rings [3].
Making use of the elasto-hydrodynamic theory, they nd that the
contact temperature on the surface of the seal ring is higher than
the temperature on the shaft surface.
In [4], the temperature distribution due to friction heat in a
radial shaft seal ring is simulated. The associated thermal deformation due to the friction power generated in the contact zone is
studied. It can be found that the inuence of thermal deformation
on the width of the contact zone and the resulting contact
pressure is small. The elastomeric behavior is modelled using
constant MooneyRivlin coefcients, independent of the current
temperature distribution. In [5] and [6], the temperature distribution within a radial shaft seal ring is calculated. The authors
conclude that the simulation of the friction torque in seal rings
requires a combined thermal and mechanical model. First, the
temperature distribution is determined and then the mechanical
behavior of the elastomeric material, taking into account its
temperature dependency, is simulated. Thereby, the contact pressure and the friction between seal and shaft can be simulated
more accurately. A detailed insight in the contact temperature of
seal rings is given in [7]. A conjugated heat transfer simulation is
conducted in order to study the inuence of the contact friction,
oil sump temperature and surrounding elements, e.g., shaft and
housing on the contact temperature. The comparison with experimental results using a thermographic camera shows a good
accordance. An empirical approach to the temperature dependence of radial shaft seal friction is presented in [8]. Through the
analysis of test rig results a correlation of friction energy and
contact temperature is deduced. In an iterative calculation, both
dimensions can be determined. Dry friction and uid friction make
up the resulting overall friction.
At the same time, the simulation of wear in seal rings is
investigated. In [9], a nite element model for the simulation of
wear in a reciprocating elastomeric seal ring under different load
conditions is published. The occurring wear is modelled in the
form of node-displacements of the nite element mesh. The effect
of wear on the resulting pressure distribution can be analyzed.

Node displacement is the rst of three common possibilities of

modelling wear in FE-models. This strategy has been adopted in
later works of other authors. In [10], wear of radial shaft seal rings
and PTFE lip seals is simulated using a similar FE-approach in the
software ANSYS. The wear is not applied as a node displacement,
but by deactivating elements. This method is called element-death
and represents the second basic possibility of modelling wear. One
advantage is that using this strategy a contortion of the mesh is
avoided. With element death, a continuous wear progress cannot
be modelled. The accuracy is limited by the element size in the
contact zone. This can be regarded as a disadvantage. The software
ABAQUS offers a FORTRAN-Subroutine umeshmotion that can be
used to move mesh-nodes independent of the underlying material
in an adaptive mesh region. This subroutine is used by [11] to
model wear in a pin-on-disc tribometer. The authors compare
experimental tribometer results with simulation results, using
Archard's wear equation and nd that the accuracy of this
modelling strategy is acceptable. A third possible modelling
strategy of wear in a FE-analysis is used by [12]. In this work,
the seal is modelled as an axisymmetric component using the
software MSC.MARC. The application of wear is not accomplished
through mesh-modications, but for each wear state, the geometry, described by contour points along the contact zone, is updated.
The current geometry is then meshed and used for further
simulation steps. This strategy is particularly powerful for wear
amounts larger than the elements in the FE-mesh.
The rst microscale models for the simulation of the tribological behavior of radial shaft seal rings have been developed
approximately 25 years ago. In 1989, Gabelli develops a model
for the simulation of the lubrication lm thickness [13]. In his
model, the lm thickness formation combines micro-EHD action,
based on interactions of surface asperities and hydrodynamic
action at the sealing interface, based on the solution of the general
Reynolds' equation. The surface roughness of seal and shaft is
described with microundulations. Salant develops a simulation
method for the estimation of the leakage rate [14], based on the
solution of the general Reynolds' equation, and taking into consideration the distortion of the seal surface undulations during
operation, as described by [15]. In regions, where the local
pressure falls below the cavitation pressure, the existence of both,
liquid and gas are taken into consideration. This simulation model
is continuously rened and developed further by Salant. Other
scientists have taken up this simulation method in their own
simulation models. Van Bavle et al. analyze different seal layouts
based on the simulation method developed by Salant [16]. Their
description of the lubrication lm is used in later research by other
scientists. In [17], the inuence of the oil meniscus on the lip seal
behavior is simulated, demonstrating, that asperity contact occurs
in the seal interface if the meniscus moves too far to the liquid
side. In [18], an extension of the model shown in [14] is presented.
It can be applied for mixed lubrication, taking into consideration
micro-cavitation and micro-deformation of the surface undulations of the seal lip. With the simulation method presented in [19],
the uid mechanics of the lubrication lm and the elastic lip
deformation are solved using the Reynolds' equation with ow
factors. This statistical approach improves the computational
efciency. This approach is extended by [20], where the deformation of the seal lip asperities is modelled, using a matrix of
elasticities that are predetermined with FE-simulations. In [21], a
thermal EHL-model of a seal ring is presented. The deformation is
also taken into account by using a matrix of elasticies. An articial
surface is used to take into consideration the surface asperities.
The system of equations is solved with the FE-method, instead of
the wide spread nite difference method.
The above listed literature illustrates the difference between
the macroscale and microscale simulation approaches. In recent

D. Frlich et al. / Wear 311 (2014) 7180

73

years, many papers that treat the temperature distribution in seal

rings, the friction torque between shaft and seal, or the wear of
seal rings separately on a macroscopic scale have been published.
In [1] and [5], rst approaches for a combined simulation of two of
these factors are presented. An approach that combines the three
aspects of contact temperature, friction and wear of seal ring and
shaft within one model is lacking so far.
Microscale models of the seal interface are typically used to
simulate lubrication lm thickness, pumping rate and leakage,
taking into consideration the inuence of deformable surface
undulations. Macroscopic modelling strategies are usually applied
for the simulation of wear during the operation of seal rings.
To reach acceptable computation times, the macroscale models of
the seal interface must be described in less detail, using empirical
approaches. The model presented in this paper falls within the
category that use macro scale approaches.

3. Model setup
The basis of the presented simulation approach is a parametric
axisymmetric nite element model created in the software
ABAQUS. The model is developed using the ABAQUS Python
scripting-interface. The seal geometry, its size as well as thermal
and mechanical boundary conditions can easily be modied in the
Python script. The denition of the simulation steps, the application of the boundary conditions and the wear application are
performed solely using Python scripting. The advantage of the
Python script is that the FE-model can easily be combined with
analytical and empirical equations, e.g., for the calculation of
boundary conditions and loads. Hence the Fortran-subroutine
umeshmotion is not used in the current model. The basic model
setup is shown in Fig. 2.
Starting with the new seal ring, the wear of the model is
simulated in a user dened number of time increments until the
requested nal wear state is reached. Within each wear increment,
three simulation steps are conducted. The rst simulation step is a
mechanical step. The seal ring geometry and the shaft geometry
are updated according to Archard's wear equation by moving the
contact nodes in the direction of the contact normal, making use
of the ALE Adaptive Mesh Constraint in ABAQUS. This displacements boundary condition on the contact nodes of seal (ds,i) and
shaft (dw,i) is illustrated in Fig. 3. With the Adaptive Mesh
Constraint, the contact nodes can be moved independently of
the underlying material. In addition, the mesh distortion in the
model areas, that are subject to wear are kept at a minimum and
acceptable element aspect ratios are maintained. Additional
boundary conditions keep the outer nodes of the seal ring metal
case and the inner nodes of the hollow shaft xed, making use of
the encastre boundary condition.
The second simulation step is a coupled temperature-displacement-step. At the beginning, contact temperature and friction
torque are calculated. It can be chosen between an empirical
approach, according to [8], and a numerical approach developed at
the MEGT, based on [22,23]. The calculated contact temperature is
applied as a temperature boundary condition on the respective
nodes of shaft (w,i) and seal (s,i) (Fig. 4). In respect to the heat
transition between the surfaces and the surrounding uids, the
heat transfer coefcients and the uid temperatures are dened as
boundary conditions (Table 1). Regarding the seal ring surface on
the oil side, fully-ooded conditions are assumed. The mechanical
boundary conditions of simulation step 1 are propagated to step 2.
Subsequently, the temperature distribution in seal ring and shaft is
determined.
In a last mechanical simulation step, the assembly of shaft and
seal ring is simulated. The boundary conditions of the previous

Fig. 2. Overview of simulation steps in the seal ring model.

Fig. 3. Boundary conditions on shaft and seal in simulation step 1 (wear step).

steps are propagated. An axial displacement is applied on the shaft

(Fig. 5). At the end of the simulation step, the rened mesh areas
of seal and shaft are in contact (Fig. 5, right).
Resulting from this step is the contact pressure, taking into
account the current wear state and the current temperature. The
contact pressure distribution of the last simulation step is the input
for the wear calculation in the subsequent wear increment. When
the nal sliding distance is reached, the simulation is aborted.

74

D. Frlich et al. / Wear 311 (2014) 7180

by hard particles between the interacting surfaces, e.g., in

lubricants.
Erosive wear results from the interaction between a solid
surface and a uid stream with a certain speed, which also
contains abrasive particles. According to [24,25], this denition
includes the impact of free-moving particles on a solid surface, too.
During the operation of machine elements such as seal rings,
the wear progress of contact surfaces can basically be divided into
three major stages. At the beginning of the rst stage, the contact
surfaces are in initial condition and subjected to a run-in. Surface
peaks are attened and in many cases friction is reduced. Often the
wear gradient is high at the beginning and declines after a short
operating time.
In the past, many scientists have investigated methods of
calculating wear. The exact number of wear equations to be found
in literature is unknown, according to some estimates, however,
there are several hundreds of them. Many of these equations can
be assigned to either a friction based approach or a contact
pressure based approach. A wide spread approach, which falls
into the second category, tracks back to Archard [26]. It is used by
many researchers for the calculation of seal wear [9,11,12].

3.1. Seal conguration

For the investigations presented in this paper, the following
seal conguration is chosen: The inner diameter of the radial shaft
seal ring is 80 mm and the outer diameter is 100 mm (Table 2).
Simplications of the geometry, e.g., the neglect of chamfers,
were conducted in areas that are assumed to have no major effect
on the simulation results. The elastomer material of the presented
seal ring consists of FKM, a uoroelastomer. Its stressstrainrelation is strongly non-linear and temperature dependent. For
the simulation of this hyper-elastic behavior, a temperature
dependent two parameter MooneyRivlin-model is chosen.
As the model is exclusively applied for the simulation of steadystate problems, the visco-elastic behavior of the elastomer material is neglected. The input parameters of the MooneyRivlinmodel have been determined by curve-tting to uniaxial stress
relaxation test data of elastomeric samples at different temperatures between 0 1C and 140 1C (Table 3).
The oil used as lubricant in the development of the thermal
model is a synthetic polyalphaolen (ISO VG 1000).

Q
V K0 s
H

3.2. Wear simulation

In this equation, V is the volumetric amount of wear, K' is the

empiric wear coefcient, Q is the contact normal force, H is the
hardness of the softer body and s the sliding distance. A modication of this equation, where the hardness is put into relation to the
wear coefcient is also very common [9,12,25]

Generally, wear describes the removal of material from a

surface, due to a mechanical load. It can be divided into four
categories, according to the occurring wear mechanism.
Adhesive wear originates from physical bonding forces like van
der Waals forces, hydrogen bridge bonds as well as metal, or ionic
bonds. In contacting bodies, the peaks of rough surfaces can
temporarily fuse together, supported by friction heat and local
pressure peaks. Due to sliding movements of the contacting
bodies, these microbonds break up and in doing so change the
surfaces.
Abrasive wear usually occurs in pairs of one hard and one soft
contact partner. The peaks of rough surfaces cut through the softer
material and as a consequence, a groove develops. This movement
is sometimes referred to as plowing. Abrasion can also be created

V KQ s

Table 1
Interaction properties of solid and uid in the coupled temperaturedisplacement-step according to [1].
Heat transfer coefcients and uid temperatures dened as interactions in simulation step 2
as

1.4

os

14.3

Heat-transfer coefcient oil-elastomer [mW2 K]

oe

81.4

Heat-transfer coefcient air-elastomer [mW2 K]

Air temperature [1C]
Oil sump temperature [1C]

ae

8.1

a
o

20
70 (depending on operating conditions)

Heat-transfer coefcient oil-steel [mW2 K]

Here, V is the volumetric amount of wear, K is the empiric wear

coefcient, Q is the contact normal force and s the sliding distance.
Eq. (2) is used for the calculation of wear in this paper. The
coefcient K is determined experimentally for the elastomer in the
seal lip and for the steal of the shaft.
In order to determine the empiric wear coefcient K, the
contour of the new seal ring and the contour of the worn seal
ring after a sliding distance of 5000 km on endurance test rigs are
overlaid (Fig. 6). The worn area between the contours is evaluated
using a Matlab-script and converted into the wear volume through
multiplication with the circumferential length of the seal.
By entering the known radial force of the seal ring and the
sliding distance of the endurance run into Eq. (2), the empiric wear
coefcient K is determined. In an earlier research project at the
MEGT institute, the inuence of shaft velocity, oil temperature,
contact pressure, and oil viscosity on the empiric wear coefcient
has been investigated experimentally [27]. A signicant relationship between oil temperature and wear coefcient has been
detected. For that reason, a temperature-dependent wear coefcient is used in the current simulation model (Table 4). For the
temperature range between 60 1C and 130 1C, the wear coefcient
is linearly interpolated. The inuence of shaft velocity, contact
pressure and oil viscosity on the wear coefcient was found to be

Fig. 4. Boundary conditions on shaft and seal in simulation step 2 (thermal step).

Heat-transfer coefcient air-steel [mW2 K]

D. Frlich et al. / Wear 311 (2014) 7180

75

Fig. 5. Boundary conditions on shaft and seal in simulation step 3 (assembly step).

Table 2
Conguration of the seal ring system.

Table 4
Empiric wear coefcients of the seal ring, determined experimentally at different
oil temperatures.

Experimental seal system conguration

Inner seal diameter [mm]
Outer seal diameter [mm]
Seal material
Shaft material
Oil

80
100
uoroelastomer
1.7131
Polyalphaolen, ISO VG 1000

Oil temperature [1C]

Empiric seal ring wear coefcient K

60
130

3.3e-6
4.5e-6

mm3
N km

Table 3
MooneyRivlin-coefcients determined from uniaxial stress relaxation test data
at different temperatures.
Temperature
(1C)

MooneyRivlin-coefcient
C01 [MPa]

MooneyRivlin-coefcient
C10 [MPa]

0
20
60
100
140

0.14
0.12
0.10
0.10
0.09

0.73
0.67
0.52
0.49
0.48

Fig. 7. Mesh of seal ring and shaft, rened in the contact zone.

wear coefcient will not be in accordance with the values presented

in Table 4.
For the application of wear in the FE-model, the denition of
adaptive mesh regions is necessary. The mesh needs to be rened
locally and the structure needs to be strictly regular, as illustrated
in Fig. 7.
3.3. Temperature calculation and friction torque

Fig. 6. The worn area (gray) is measured by overlaying microscopic images of the
new seal ring contour and the contour of the worn seal after 5000 km sliding
distance on the test rig (schematic).

less strong in [27]. Thus it is neglected in the simulation. Experimental results of the current seal setup have shown no signicant
wear of the shaft. Therefore the empiric wear coefcient of the
shaft is assumed as 0 mm3/N km. It must be noted that in seal
ring applications, where oil viscosity, elastomer material, or seal
geometry differ considerably from the presented setup, the empiric

Due to solid friction between seal ring and shaft and due to
uid lm friction, the temperature in the seal system has a local
maximum in the contact zone. This temperature maximum has a
strong inuence on the elastic properties of the elastomeric
material. A simplied approach for the calculation of friction
power in radial shaft seal rings is proposed by [28].
P f pvdb

Here P is the friction power, f is the coefcient of friction, P is the

contact pressure, v is the circumferential velocity of the shaft, d is
the shaft diameter, and b is the contact width.
In recent works, more complex and more accurate approaches
for the calculation of friction power and rst approaches for the
calculation of contact temperature have been published.
In [8], an empiric approach is presented. Based on a large
number of experiments, a friction model is developed. The friction

76

D. Frlich et al. / Wear 311 (2014) 7180

Table 5
Parameters of PAO and reference oils according to [32,33].
Oil parameter

PAO

Ref. Oil1

Ref. Oil2

Ref. Oil3

Ref. Oil4

viscosity class ISO VG

Vogel parameter A [Pas]

1000

15

32

100

460

5:588922  10  5
5715.756
300
0.316144

8:22563  10  5
693.2438
95
0.6629436

7:0840  10  5
800.3
95
0.8255

2:84477  10  5
1235.573
115
0.5490735

1:37492  10  5
1696.304
125
0.4977719

 8:661369  10  3
1400

1:22344  10  2
2000

 2:31  10  2
2000

8:88301  10  3
2050

1:82924  10  3
2000

Vogel parameter B [1C]

Vogel parameter C [1C]
Rodermund parameter D []
Rodermund parameter E []
Rodermund parameter F [bar]

torque in this model is constituted of a non-viscous fraction at zero

speed and a velocity-dependent fraction.
 3
d
b
d 22
n
T T 0 T 0 F r Rp
2

2
60

Here T is the total friction torque, T0 is the initial friction torque,

T is the viscous part of the friction torque, 0 is the boundary
friction coefcient, Fr is the radial force, d is the shaft diameter, is
the dynamic uid lm viscosity, is the contact temperature, b is
the contact width, Rp is the sum of the surface roughness Rp of
shaft and seal lip, and n is the rotational speed of the shaft.
The temperature in Eq. (4) can be calculated as the sum of oil
temperature oil and over temperature c in the contact area. The
over temperature is dependent on the friction power P per area A.
As a result of measurements with different seal ring materials,
radial forces and lubricants, [8] suggests an over temperature of
16.5 K in the contact area for a friction power per area of 1 W/mm2.
 
P
5
oil o ; where o f
A

uid velocity. For the calculation of the temperature prole of the

lubricant, a parabolic distribution is assumed.
The pressure and temperature dependence of the uid lm
viscosity is determined by using the Rodermund equation [31]


D EB=C
; p
B p  p0
ln

1
8
A
C
F
The oil-parameters that were used for the viscosity model are
listed in Table 5.
The mean viscosity m can be determined as the arithmetic
mean of the viscosity across the uid lm:
Z hf ilm
1
m
; pdy
9
hf ilm 0
As the oil in the contact zone between seal ring and shaft is
subject to considerable shear, Eyring's material law [34] is used to
take into account the non-Newtonian behavior
m

The model presented in this paper solves Eqs. (4) and (5) in a
recursion. The resulting contact temperature is applied on the
contact nodes of shaft and seal lip as a boundary condition.
It should be noted that the experimental determination of the
boundary friction coefcient 0 is of great importance. The method
of determination is described in detail in [8]. In addition it is
shown, that for a wide range of oils and elastomers, 0 can be
assumed as 0.3 in a good approximation. Therefore the value
0 0.3 is used in the presented simulation model.
A second option for the contact temperature calculation in the
model, presented in this paper, is a numerical contact temperature
calculation, based on [22,23]. The basis of this method is the
energy balance, taking into account the heat generated by internal
friction as well as the heat transport towards the wall. As a rst
step, the equations for the surface temperatures of the contacting
bodies can be written as a simplied form of Fourier's law for heat
conduction [29,30]:
 2

ci i vi i
i 2 i

2 0
6
2
i x
x
y
The energy equation implemented in the seal ring model
presented in this paper has the following form:
c1

1
2
v2
c2
x 0
y1
y2
h0

In this equation c1 and c2 are the heat conductivity of the

contacting bodies, 1 =y1 and 2 =y2 are the temperature
gradient on the surface of the two contacting bodies, while is
the dynamic uid lm viscosity and x is the x-component of the

=Ey
sin h=Ey

10

is the oil lm shearing, and Ey is the Eyring shear stress, that

is the critical stress where Newtonian uid behavior changes to
non-Newtonian.

4. Simulation results
Regarding the simulation results, it can be distinguished
between rst, the analysis of the new seal ring in its initial
condition, second, the analysis of the seal system during endurance operation, and, third, the analysis of the worn seal ring and
shaft after operation.
4.1. Initial condition of the seal system
Wear, friction and contact temperature are signicantly inuenced by the initial contact width and the initial contact pressure
between shaft and seal lip. The lip contour in the FE-model needs
to be modelled as close to the real seal lip as possible. Therefore
the seal is analyzed using a fringe projection microscope (Fig. 8).
A cut across the seal lip reveals a seal lip radius of approximately
215 mm. This radius is used to dene the lip geometry of the
FE-model.
In order to check the accuracy of elastomer stiffness and lip
geometry, preliminary tests of contact width and radial force are
conducted. The contact width of seal ring and shaft is analyzed by
mounting the seal ring on a hollow acrylic glass shaft. The contact
area is visualized using a microscope. In the FE-model, the contact
width is determined by measuring the distance of the outermost
nodes that show a positive contact pressure. It can be shown that

D. Frlich et al. / Wear 311 (2014) 7180

77

Fig. 8. Fringe projection microscope picture of a FKM seal lip, with a lip radius of approximately 215 mm.

Fig. 9. Microscopic picture of the initial contact width of a seal ring on a hollow glass shaft (left), and the initial contact width in the nite element model (right).

Fig. 10. Schematic depiction of the shaft, divided in two halves for the measurement of radial force (left), comparison of initial radial force at different seal ring temperatures
in experiment and simulation (right).

experimental and simulated contact width are in good accordance

(Fig. 9).
The radial force of seal ring and shaft is determined experimentally by using a test rig according to [35] (Fig. 10). The seal ring
is mounted on a split shaft. One half of the shaft is xed on a beam
in bending. When mounted, the seal ring reduces the gap between
the two halves. This reduction is measured and converted into the
radial force.
As the radial force is highly temperature dependent, the measurement is conducted at several temperatures over the anticipated
operating conditions, from room temperature at 25 1C up to a
maximum temperature of 190 1C. Fig. 10 shows that the accordance
of simulated and measured radial forces is very good.

4.2. Temperature and friction during operation

During endurance operations, the simulation model offers a
wide range of possibilities to analyze the contact temperature and

the temperature distribution of the seal system in different states

of wear. The following section demonstrates for a new seal ring
the resulting contact temperature at different working conditions.
At an oil sump temperature of 80 1C, the contact temperature
rises from 93 1C to 164 1C when the rotational speed of the shaft is
increased from 500 min  1 to 3000 min  1 (Fig. 11). For a seal ring
diameter of 80 mm, the corresponding range of surface velocities
is 2.1 m/s to 12.6 m/s. The lubricant used is a synthetic polyalphaolene. The simulated contact temperature is in good accordance in the empirical and numerical simulation approach.
When the rotational speed is kept constant at 2000 min  1 and
the oil sump temperature is increased from 40 1C to 100 1C the
maximum contact temperature rises from 110 1C to 150 1C (Fig. 12).
At a constant rotational speed of 2000 min  1 and an oil sump
temperature of 80 1C, the contact temperature shows a dependency on the oil viscosity (Fig. 13).
For lubricants with a lower viscosity (ref1, ref2, ref3), the
contact temperature amounts to 125 1C, while for lubricants with
a higher viscosity, the contact temperature rises to approximately

78

D. Frlich et al. / Wear 311 (2014) 7180

4.3. Wear of the seal system during operation

During the operation of the seal system, an advancing material
removal of the seal lip can be detected. Due to the removal and the
connected contour change of the seal lip, the tribological situation
in the contact area changes permanently. The different aspects,
which are affected by wear, can be analyzed with the presented
simulation model in detail. The simulated change of the seal lip
contour up to a maximum sliding distance of 5000 km at a
rotational speed of 3000 min  1 is illustrated in Fig. 15.

Fig. 11. Contact temperature for different rotational speeds of the shaft calculated
with the empiric and the numeric model.

Fig. 14. Temperature distribution along the contact width on the seal ring surface,
calculated with the numeric model.

Fig. 12. Contact temperature for different oil sump temperatures calculated with
the empiric and the numeric model.

Fig. 15. Simulated seal lip contour at different states of wear at a shaft rotational
speed of 3000 min  1, starting with the new seal at 0 km and resulting in the worn
seal at 5000 km.

Fig. 13. Contact temperature for different lubricants calculated with the empiric
and the numeric model.

140 1C. Again, the empirical and numerical simulation approaches

are in good accordance.
In addition to the maximum over temperature in the seal
contact, the numerical simulation approach provides the distribution of the contact temperature along the contact width of the seal
lip (Fig. 14). It can be illustrated, that the contact temperature
distribution is similar to the distribution of contact pressure.
At higher rotational speed, the difference between the temperature in the middle of the contact zone and the temperature at the
edge of the contact zone becomes more distinctive.

Fig. 16. Asymmetric contact pressure distribution for different states of wear.

D. Frlich et al. / Wear 311 (2014) 7180

It can be seen that the initial contour of the seal lip is modelled
as a radius. The asymmetric contact angles of the seal ring
inuence the position of the material removal. The contact angle
on the air side is atter than the angle on the oil side, which is why
the area of occurring wear is shifted to the air side. The change of
the lip contour affects the contact pressure distribution in the
simulation (Fig. 16). The maximum contact pressure is reduced
from 1.8 MPa at the beginning to 1.1 MPa at 5000 km.
The radial force as an integral of the contact pressure over the
contact area is illustrated in Fig. 17. It can be noted, that even though
the pressure distribution changes signicantly during the operation of

Fig. 17. Simulation results for the development of wear volume and radial force
during seal operation.

Fig. 18. Simulated friction torque and contact width development during the seal
operation.

Fig. 19. Friction torque development during seal operation, based on experimental
results in [36].

79

the seal ring, the radial force is reduced by just 2%. Since the load in
the wear Eq. (2) does not change signicantly during operation, the
simulated wear volume development during operation is approximately linear in our simulation model (Fig. 17). In experiments the
wear volume will show a greater slope at the beginning that decreases
towards the end of the operation time. This running-in is not taken
into account due to the use of Archard's wear equation.
At the same time, the contact width grows from 100 mm to
approximately 180 mm at 5000 km (Fig. 18). The simulation results
reveal an increase of the friction torque T during operation (Fig. 18).
The cause for the increasing friction torque can be found in the
superposition of two opposing effects. On the one hand, the radial
force decreases slightly, due to the material removal. The non-viscous
component of the friction T0 (Eq. (4)) decreases with the radial force
as well. On the other hand, a growing amount of lubricant is subject
to shear in the contact zone, due to the increasing contact width. This
leads to an increase of the viscous component of the friction torque
T (Eq. (4)). The inuence of the growing contact width is dominant
and the resulting overall friction torque T increases.
The development of friction torque during endurance seal ring
operations has also been investigated by [36]. For a comparable
seal ring system at constant temperatures and a constant shaft
velocity, it could be found experimentally that the friction torque
increases over operation time (Fig. 19) which conrms the simulated friction torque development in Fig. 18.
The increasing friction torque leads to an increase of the
friction power during the seal operation (Fig. 20). Due to the
increase of the contact width the resulting friction power per
contact area decreases, as illustrated in Fig. 20.

Fig. 20. Simulated development of friction power and friction power per contact
area during the seal operation.

Fig. 21. Contact temperature distribution for different states of wear.

80

D. Frlich et al. / Wear 311 (2014) 7180

Fig. 22. Contact temperature development during seal operation, calculated with
the empiric and the numeric model.

The friction power per contact area is the signicant dimension

that inuences the temperature in the contact zone of seal ring and
shaft. The simulated temperature distribution correlates with the
decreasing friction power per area and has a distinct peak of 125 1C.
Initially the contact temperature is distributed over a small contact
width (Fig. 21). With progressing wear, the contact temperature is
distributed over a wider contact zone and shows a maximum contact
temperature of 113 1C at the end of the simulation.
For one spot in the center of the contact zone, the contact
temperature development shows a very good correlation when
comparing empirical and numerical simulation models (Fig. 22).
5. Conclusion
In this paper, a comprehensive simulation approach for radial
shaft seal rings is presented. The model can be applied to analyze
several macroscopic aspects in the system of radial shaft seal rings.
Wear of radial shaft seal rings can be simulated with good
accuracy as the temperature-dependent elastomer stiffness is taken
into consideration. The detailed insight into the contact temperature
distribution helps to choose an appropriate elastomeric material.
Especially the formation of oil carbon due to high contact temperatures can thus be avoided. Also the inuence of wear on the friction
torque can be analyzed. With the parametric model setup design
optimizations, e.g., minimizing the friction torque can be conducted.
It must be noted, however, that the equations for the simulation of
wear and friction are based on empirical approaches. Thus the scope
of application of the listed friction and wear coefcients is limited to
conditions similar to those presented in this paper. Moreover, it is
important to note that the local distribution of temperature and
pressure on a microscopic scale is inuenced by surface asperities.
With the presented simulation approach these local pressure and
temperature changes cannot be taken into account.
Acknowledgments
The authors would like to thank the German Research Foundation (DFG) for the support of the research within the Collaborative
Research Centre 926 Microscale Morphology of Component
Surfaces (MICOS), sub-project C01.
References
[1] G. Upper, Dichtlippentemperatur von Radialwellendichtringen (Ph.D. thesis),
Karlsruhe, 1968.

[2] M.J.L. Stakenborg, R.A.J. Ostayen, Radial lip seals, thermal aspects, Tribol. Des.
Mach. Elem. 14 (1989) 7988.
[3] Y.S. Kang, F. Sadeghi, Numerical analysis of temperature distribution at the lip
seal-shaft interface, J. Tribol. 119 (1997) 273278.
[4] C. Lee, C. Lin, R. Jian, C. Wen, Simulation and experimentation on the contact
width and pressure distribution of lip seals, Tribol. Int. 39 (2006) 915920.
[5] S. Meyer, O. von Estorff, D. Feldmann, M. Wollesen, Test on Rotary, Shaft Lip
Type Seals and Finite Element Simulation for Validation of, Material Laws for
Elastomers, 14th ISC Stuttgart, 2006, pp. 522533.
[6] S. Plath, S. Meyer, V. Wollesen, Friction torque of a rotary shaft seal lip type
seal a comparison between test results and nite element simulation, Mech.
Kaunas. (2005) 5559.
[7] A. Daubner, W. Haas, Simulation der Temperatur im und am Reibkontakt,
Computational Fluid Dynamics, NAFEMS Mag. (2009) 6271.
[8] T. Engelke, Einuss der Elastomer-Schmierstoff-Kombination auf das, Betriebsverhalten von Radialwellendichtringen (Ph.D. thesis), University of Hannover,
2011.
[9] C. Debler, Bestimmung und Vorhersage des Verschleies fr die Auslegung
von Dichtungen (Ph.D. thesis), University of Hannover, 2005.
[10] Z. Wang, D. Draper, T. Hodapp, Radial Lip Seal Simulation Using ANSYS NonStandard Procedures, International ANSYS Conference, Pittsburgh, USA, 2006,
pp. 112.
[11] V. Hegadekatte, S. Kurzenhuser, N. Huber, O. Kraft, A predictive modeling
scheme for wear in tribometers, Tribol. Int. 41 (11) (2008) 10201031.
[12] N. Bksi, K. Vradi, Wear Simulation of a Reciprocating Seal by Global
Remeshing, Periodica Polytechnica (Mechanical Engineering), Budapest
(2010) 7175.
[13] A. Gabelli, Micro-elastohydrodynamic lubricant lm formation in rotary lip
seal contacts, in: Proceedings of the 15th Leeds-Lyon Symposium on Tribology,
Tribology Series, vol. 14, 1989, pp. 5768.
[14] R.F. Salant, Numerical analysis of the ow eld within lip seals containing
microundulations, J. Tribol. 114 (3) (1992) 485491.
[15] M. Kammler, Zur Abdichtwirkung von Radial-Wellendichtringen (Ph.D.
thesis), Universitt Stuttgart, 1986.
[16] P.G.M. Van Bavel, T.A.M. Ruijl, H.J. Van Leeuwen, E.A. Muijderman, Upstream
pumping of radial lip seals by tangentially deforming rough seal surfaces, J.
Tribol. 118 (2) (1996) 266275.
[17] R.F. Salant, Modelling rotary lip seals, Wear 207 (12) (1997) 9299.
[18] F. Shi, R.F. Salant, A mixed soft elastohydrodynamic lubrication model with
interasperity cavitation and surface shear deformation, J. Tribol. 122 (1) (2000)
308316.
[19] R.F. Salant, A.H. Rocke, Hydrodynamic analysis of the ow in a rotary lip seal
using ow factors, J. Tribol. 126 (1) (2004) 156161.
[20] F. Guo, X. Jia, S. Suo, R.F. Salant, Y. Wang, A mixed lubrication model of a rotary
lip seal using ow factors, Tribol. Int. 57 (1) (2013) 195201.
[21] M. Hajjam, D. Bonneau, Inuence of the roughness model on the thermoelastohydrodynamic performances of lip seals, Tribol. Int. 39 (3) (2006)
198205.
[22] B. Magyar, Tribo-dynamische Untersuchungen von Zylinderschneckengetrieben (Ph.D. thesis), University of Kaiserslautern, 2012.
[23] B. Magyar, B. Sauer, P. Hork, Tribological investigation of K type worm gear
drives, Acta Polytechnica Hungarica 9 (6) (2012) 233252.
[24] D. Felhs, Dry sliding and rolling tribotests of carbon black lled EPDM
elastomers and their fe simulations (Ph.D. thesis), University of Kaiserslautern,
2008.
[25] S.W. Zhan, Tribology of Elastomers, Elsevier, Amsterdam, 2004.
[26] J.F. Archard, Contact and rubbing of at surfaces, J. Appl. Phys. 24 (8) (1953)
981988.
[27] T. Gastauer, Reibungs- und Verschleiuntersuchungen von Elastomerwerkstoffen fr Dichtungsanwendungen (Ph.D. thesis), University of Kaiserslautern,
2013.
[28] H. Mller, Abdichtung bewegter Maschinenteile, Funktion-Gestaltung-Berechnung-Anwendung, Waiblingen, Medienverlag U. Mller, 1990.
[29] H.S. Carslaw, J.C. Jaeger, Conduction of Heat in Solids, Oxford University, Press,
Oxford, 1959.
[30] H. Plote, Zur Berechnung thermo-elasto-hydrodynamischer Kontakte (Ph.D.
thesis), University of Clausthal, 1997.
[31] H. Rodermund, Extrapolierende Berechnung des Viskosittsverlaufes unter
hohen Drcken, Tribol. Schmierungstech. 27 (1980) 35.
[32] G. Eller, Ein Beitrag zur Berechnung des stationren nichtisothermen Elastohydrodynamischen Schmierlms (Ph.D. thesis), University of Karlsruhe, 1987.
[33] J. Kroll, G. Noack, Zur Bestimmung der Viskositt von binren lmischungen
unter, hohen Drcken/Deutsche Wissenschaftliche Gesellschaft fr Erdl,
Erdgas und Kohle, E.V. (Forschungsbericht 451). Research Report, , 1993.
[34] B.J. Hamrock, S.R. Schmid, B.O. Jacobson, Fundamentals of Fluid Film Lubrication, second ed., Marcel Dekker Inc., 2004.
[35] DIN 3761 T9, Radial-Wellendichtringe fr Kraftfahrzeuge Prfung Radialkraft- Megert-Digital, Deutsche Norm, Berlin, 1984.
[36] M. Wilke, O. von Estorff, V. Wollesen, Development of a Dynamical Test
Method to Evaluate Combinations of Oil and Elastomers with Respect to
Friction and Abrasion, DGMK-Research ReportUniversity Hamburg-Harburg,
2011.