ArticlePDF Available

# Sensitivity of the Stribeck curve to the pin geometry of a pin-on-disc tribometer

Authors:

## Abstract and Figures

A considerable number of surface texture investigations is based on pin-on-disc tribometers. This work shows that a crucial role in the reproducibility of the results, e.g. Stribeck curves, is played by the geometry of the pin surface. The investigation is based on an elastohydrodynamic model of a pin-on-disc tribometer which is validated with experimental data. Characteristic roughness and pin shapes are introduced in this model to evaluate the sensitivity of the Stribeck curve to these operating conditions. The obtained significant variations in the friction coefficient indicate that studies which aim at quantifying the influence of surface textures in the mixed lubrication regime need to provide information about the pin geometry in order to enable a meaningful comparison among literature data.
Content may be subject to copyright.
Sensitivity of the Stribeck curve to the pin geometry of
a pin-on-disc tribometerI
Erik Hansen
Institute of Fluid Mechanics, Karlsruhe Institute of Technology (KIT),
Kaiserstr. 10, 76131 Karlsruhe, Germany
Bettina Frohnapfel
Institute of Fluid Mechanics, Karlsruhe Institute of Technology (KIT),
Kaiserstr. 10, 76131 Karlsruhe, Germany
Andrea Codrignani
Department of Microsystems Engineering, Albert-Ludwigs-University Freiburg,
Georges-K¨ohler-Allee 103, 79110 Freiburg, Germany
Abstract
A considerable number of surface texture investigations is based on pin-on-
disc tribometers. This work shows that a crucial role in the reproducibilty of
the results, i.e. Stribeck curves, is played by the geometry of the pin surface.
The investigation is based on an elastohydrodynamic model of a pin-on-disc
tribometer which is validated with experimental data. Characteristic rough-
ness and pin shapes are introduced in this model to evaluate the sensitivity
of the Stribeck curve to these operating conditions. The obtained signiﬁ-
cant variations in the friction coeﬃcient indicate that studies which aim at
quantifying the inﬂuence of surface textures in the mixed lubrication regime
need to provide information about the pin geometry in order to enable a
meaningful comparison among literature data.
Keywords: mixed lubrication, modelling in tribology, pin-on-disc
IThis document is the result of a research project funded by the Friedrich und Elisabeth
Boysen-Foundation.
Corresponding author
URL: www.istm.kit.edu (Erik Hansen)
Preprint submitted to Tribology International October 21, 2019
tribometer, surface texturing
1. Introduction
Since 20% of the world’s total energy consumption goes to overcome fric-
tion [1], research on drag reduction bears great potential in energy saving.
One of these technologies is the introduction of surface textures in lubricated
contacts. Surface textures in the shape of dimples can cause an additional
pressure build-up within the lubrication ﬂow, which leads to a thicker ﬂuid
ﬁlm and less friction [2]. This positive eﬀect is highly sensitive to the con-
tact’s operating condition and the robust and versatile design is a subject of
intensive research [3, 4]. Due to the high sensitivity of the texture parameters,
the operating conditions during their investigations must be known and con-
trolled. A widely used way to investigate surface textures under such isolated
conditions is the pin-on-disc tribometer [5]. The knowledge of all operating
parameters in this experiment allows its numerical representation, while the
experimental results enable the meaningful validation of the created digital
twin. Among others, such studies were previously carried out in the context
of a thrust bearing [6] or a piston-ring-liner contact [7]. In the case of a pin-
on-disc tribometer, setting up a digital twin based on previously published
experimental data proves to be diﬃcult because important information, such
as a measurement of the pin curvature, is often missing [8, 9, 10, 11, 12, 13].
For the present study additional data of the pin geometry employed in the
experiments of Braun et al. [8] is used in order to determine the sensitivity
of numerical predictions on this quantity. In consequence, a suggestion of the
surface parameters that should be published along with experimental results
The paper is structured as follows. The reference pin-on-disc tribometer
is introduced in section 2 before the setup of the corresponding numerical
model is explained in section 3. This model is based on a two-scale approach
in which the roughness scale is considered through precalculated contact
pressure and homogenization factors [14] while the modelling of the elas-
tohydrodynamic lubrication (EHL) is performed through the homogenized
Reynolds equation including mass-conserving cavitation [5, 15, 16] and the
boundary element method [17]. In section 4, the implemented model is used
to quantify the inﬂuence of the pin geometry on the Stribeck curve.
2
2. Reference Tribometer set-up
A schematic setup of the Plint TE-92 HS tribometer from Phoenix Tri-
bology (Kingsclere, UK) that was used for the experiments of Braun et al.
[8] is depicted in Figure 1. It shows the rotating disc that is pressed with the
normal force FNagainst the pin. The pin is placed on a self alingning pin
holder and the contact of pin and disc is constantly fed with oil. The setup
is heated to keep the oil temperature at 100C. At this temperature, the ad-
ditive free poly-alpha-oleﬁn (PAO), Kl¨uber Lubrication (Munich, Germany)
has a dynamic viscosity of approximately µl= 0.0031Pa ·s. The pin with
a diameter of 8mm consists of normalized steel C85 (Stahlbecker, Heusen-
stamm, Germany) with a Young’s modulus of approximately E= 206GPa
and a hardness of 400HV. The disc with a diameter of 70mm is made out
of hardened and tempered (190C) steel 100Cr6 (AISI 5210, Eisen Schmitt,
Karlsruhe, Germany) with a hardness of 800HV.
Figure 1: Schematic set-up of the pin-on-disc tribometer as presented in [8].
During the experiments, the friction force FTacting on the pin surface is
measured to determine the friction coeﬃcient Cf:
Cf=FT
FN
. (1)
The Stribeck curve is measured for diﬀerent disc velocites Uby approx-
imating them as U= ·rs, where Ω is the angular velocity of the disc and
3
rs= 30mm is the distance from the pin center to the rotation center of the
disc. Consequently, this approximation neglects velocity gradient eﬀects [18].
In order to numerically reproduce the experimental Stribeck curve, the
forces FNand FTon the pin surface Amust be computed. This is performed
by superposition of the hydrodynamic, ambient and contact pressures, phd,
paand pc, and the corresponding shear stresses, τhd and τc[5]:
FN=ZA
ptotdA=ZA
(phd pa+pc) dA, (2)
FT=ZA
τtotdA=ZA
(τhd +τc) dA. (3)
Realistic values for these two quantities need to be determined in order to
study the inﬂuence of pin surface geometry and pin roughness in the digi-
tal twin of the experiment. Unfortunately, this information is typically not
provided in literature.
As reference for the present study we complement the results of Braun
et al. [8] with additional measurement data that are accessible. For the
macroscopic pin proﬁle, a surface measurement along the centerline of an
unused pin is considered and extrapolated. The result is presented in Figure 2
where rounding of the pin surface is clearly visible. This macroscopic surface
geometry originates from the polishing step during manufacturing. It should
be noted that this pin surface has not yet been subjected to wear and running
in eﬀects which are likely to generate geometrical changes. Therefore, the
available surface geometry can only serve as a rough estimate of the maximum
gap height variations across the pin.
In contrast to the macroscopic pin geometry, information about the pin
roughness is available for used pins. The roughness proﬁle on a run-in pin
was measured with an optical 3D-proﬁlometer of the type PLu from SensoFar
(Barcelona, Spain) in an interferometric measuring mode with a ten times
magnifying lens. The roughness proﬁle is shown in Figure 3 with the mean
plane set to 0µm. The center-line average is Ra= 0.107µm, the standard
deviation σ= 0.143µm, the skewness Sk =1.26, the kurtosis Ku = 4.50,
the maximum of Rp= 0.287µm and the minimum of Rv=0.656µm, where
these values were calculated according to Bhushan [19].
4
Figure 2: Macroscopic pin proﬁle with
curvature due to the polishing proce-
dure. Note that the x3-axis is more
than 1000 times magniﬁed in order to
outline the curvature of the pin.
Figure 3: Representative roughness
patch with grooves due to manufactur-
ing and wear.
3. Numerical approach
3.1. Fluid mechanics
Based on the model of Wolozynski et al. [16] in the implementation of
Codrignani et al. [5], the hydrodynamic pressure distribution phd in the gap
of height hbetween pin and disc is described by the Reynolds equation con-
sidering a mass-conserving cavitation algorithm and the cavitation condition:
∇ · h3phd 6lU
0(1 θ)= 0, (4)
(phd pcav)θ= 0. (5)
The cavitation pressure is estimated to pcav = 0.08MPa following [20].
The cavity fraction θ= 1 ρ
ρldescribes the amount of cavitated lubricant,
where ρlis the density of the liquid lubricant and ρis the density of the
mixture phase. This equation system is discretized with the ﬁnite-volume-
method and solved using the Dirichlet boundary condition of ambient pres-
sure pa= 101325Pa and no cavitation at the domain boundaries. The corre-
sponding hydrodynamic shear stress τhd on the pin surface is determined as
5
[7]:
τhd =h
2
∂phd
∂x1
+µlU
h(1 θ). (6)
In order to consider roughness eﬀects on the pin, the gap coordinate h0
is introduced. It descibes the distance of the disc to the mean plane of the
undeformed roughness proﬁle on the pin. The disc is assumed to be perfectly
ﬂat. For a value of h0Rp= 0.287µm, surface contact between the disc
and the roughness proﬁle occurs. Since this causes a deformation of the
roughness proﬁle, h0is not an adequate description of the mean gap height
during surface contact. Therefore, in analogy to Forooghi et al. [21], the
meltdown gap height hmis deﬁned as the distance of the disc to the mean
plane of the deformed roughness proﬁle. It is equal to h0for h0> Rpbut
diﬀers during surface contact. Additionally, it is assumed that the surface
contact of disc and roughness proﬁle cuts oﬀ roughness asperities but leaves
a remaining gap height of = 108m to improve numerical stability. This is
schematically displayed in Figure 4 and the resulting meltdown gap height
as a function of the gap coordinate h0is shown in Figure 5. Using this
dependency, the gap height hin equation (4) can be approximated by the
roughness meltdown gap height hmas a function of the gap coordinate h0:
h=hm(h0). (7)
6
h0Rp
h0hm
h0> Rp:
h0< Rp:
hm
Figure 4: Schematic comparison of gap
coordinate and meltdown gap height
with and without surface contact.
h0=Rp
Figure 5: Meltdown gap height as a
function of the gap coordinate during
surface contact for the speciﬁc rough-
ness shown in Figure 3.
While using equation (7) as a description of the gap height in the Reynolds
equation (4) only allows to take the mean plane of a roughness proﬁle into
account, the homogenization of the Reynolds equation enables the considera-
tion of the roughness’s general shape and orientation [15]. This approach de-
composes the gap height hinto the meltdown gap height hmand a roughness
gap height hr, which depend on macroscopic coordinates ~x and roughness
coordinates ~
ξ:
h=h(~x, ~
ξ) = hm(~x) + hr(~x, ~
ξ). (8)
It is assumed that hris periodic over the representative roughness domain
A~
ξwith the lengths Y1and Y2and that these lengths are signiﬁcantly smaller
than any characteristic length in the macroscopic coordinates ~x. In this
case, certain terms can be neglected when equation (8) is substituted into
the Reynolds equation (4), which is after an asymptotic expansion of the
hydrodynamic pressure phd averaged over the periodic domain A~
ξ. The result
is the homogenized Reynolds equation and the analogously derived expression
for the homogenized shear stress in dependence of the homogenization factors
A,~
b,~c and d[15, 22]:
∇ · Aphd +~
b(1 θ)= 0, (9)
τhd =~c · ∇phd +d(1 θ) . (10)
7
Summarizing, the homogenization approach for the representation of the
microscale roughness is as follows: during the preprocessing, the local prob-
lems summarized in Appendix A are solved for various imposed gap co-
ordinates h0and the employed roughness proﬁle under consideration of a
minimum remaining gap height of = 108m in between roughness and disc.
From their solutions, the homogenization factors are computed as a function
of the gap coordinate. This dependency of the normalized homogenization
factors is displayed in Figure 6. Since the homogenization factors are normal-
ized with the meltdown gap height hm, the homogenization method coincides
with the concept of a hydrodynamically smooth surface along the roughness
mean plane as long as the normalizations of A11,A22,b1are equal to 1 and
A12,A21,b2are equal to 0. Otherwise, the ﬂow factors of both methods diﬀer
from each other, which becomes visible for values below h0= 1µm.
While the microscale is represented through the homogenization approach
described above, the macroscopic pin geometry is taken into account through
the assignment of a gap coordinate h0to each coordinate ~x. This allows
solving the homogenized Reynolds equation on the macroscopic scale during
the main processing.
Figure 6: Normalized homogenization factors as function of the gap coordinate for the
speciﬁc roughness shown in Figure 3.
8
3.2. Contact mechanics
In analogy to the homogenization factors, the contact pressures and shear
stresses are determined during the preprocessing as a function of the gap co-
ordinate h0and the employed roughness proﬁle. For each imposed h0, the
real area of contact Ac(h0) between the roughness proﬁle and the smooth disc
is determined. Note that at this point, the earlier considered minimum re-
maining gap height of = 108m is not necessary for any numerical stability
and therefore not employed. Tabor [23] states that the contact of two rough
bodies does not actually occur on their whole macroscopic surface, but only
on a small fraction of it called the real area of contact. This area is described
by the contact of the surface asperities which are almost immediately plasti-
cally deformed until the real area of contact is large enough to support the
whole contact load. Since almost the entire real area of contact is plastically
deformed, the occuring contact pressures can be linked to the material’s yield
strength σY. If it is further assumed that the asperites are of a small height
and there is no relative movement between the surfaces, the contact pressure
pcin a macroscopic discretization cell of area Atot is described as:
pc= 3σY·Ac(h0)
Atot
. (11)
The precalculated contact pressure in depence of the gap coordinate is
visualized in Figure 7. Bowden and Tabor [24] also give an expression for
the shear stress τcin the contact surface of two metals without a normal
load. They explain that the metals form welded bridges which have to be
sheared oﬀ by relative motion. By applying the Mises criterion of equivalent
stress, the shear stress can be calculated as a function of the material’s yield
strength:
τc=σY
3·Ac(h0)
Atot
(12)
Assuming ideal elastic-plastic material properties, the yield strength can
be approximated from the Vickers hardness as described in Appendix B.
It is important to mention that equation (11) assumes no shear load while
equation (12) assumes no normal load. If both are present, contact and shear
stresses are actually lower than those given by the equations. However, these
equations are used in the employed model since they excel in computation
eﬃciency and the resulting friction coeﬃcent in the boundary lubrication
regime reﬂects experimental data reasonably well.
9
Figure 7: Contact pressure as a function of the gap coordinate. The value of h0= 0
corresponds to the roughness meltdown plane for the speciﬁc roughness shown in Figure
3.
3.3. Elastohydrodynamic deformation
The sum ptot of the hydrodynamic and contact pressure ﬁelds elastically
deforms the pin on the macroscopic scale, thereby changing the initial gap
height distribution. To take this into account, the elastic deformation hel of
the pin surface Ais described by the elastic half-space model [17]:
hel(x1, x2) = (1 ν2)
πE ZZ
A
ptot(x0
1, x0
2)
q(x1x0
1)2+ (x2x0
2)2dx0
1dx0
2, (13)
where νis Poisson’s ratio and Eis Young’s modulus. When the surface A
is discretized into rectangles, equation (13) can be expressed in dependency
of a Kernel function K[17, 25]:
hel(x1, x2) = X
x0
1X
x0
2
K(x1x0
1, x2x0
2)·ptot(x0
1, x0
2). (14)
The computation of the elastic deformation due to a pressure ﬁeld is
accelerated by using the Fourier transformation F[17]:
hel =F1(F(K)· F (ptot )) . (15)
10
On the downside, at this point the discretized domain has to be increased
and padded with zeros to perform a linear convolution instead of a circular
one, which increases the computational costs of the convolution. After its
calculation, the domain is resized to its old range. The resulting gap coordi-
nate h0is then computed as the superposition of the gap coordinate above
the initially rigid pin h0,ri and its elastic deformation hel:
h0(ptot) = h0,ri +hel (ptot ). (16)
Since h0depends upon the total pressure proﬁle ptot , which also depends
upon the gap coordinate h0, ﬁnding the equilibrium requires an iterative
procedure. At ﬁrst, for an initial pressure proﬁle p(i)
tot, the elastic displace-
ment and its resulting pressure distribution ptot,II are computed. Then, the
residuum rel of the two pressure ﬁelds is calculated and the pressure ﬁeld for
the next iteration step p(i+1)
tot is determined by underrelaxation as long as rel
is higher than a threshold of tol = 105:
rel =1
NpZNp
|ptot,II p(i)
tot|
pa
dn, (17)
p(i+1)
tot =p(i)
tot +α(ptot,II p(i)
tot). (18)
where Npis the total number of the discretization points nabove the
pin, pais the ambient pressure and αis the underrelaxation factor. It is set
to α= 0.5 in the hydrodynamic and α= 0.05 in the mixed and boundary
lubrication regime to achieve a good tradeoﬀ between convergence speed and
stability.
4. Results
A digital twin of a pin-on-disc tribometer should eventually allow the
prediction of Stribeck curves. While exact agreement with experimental
data is challenging [5], the present model can be used to indicate poten-
tial sensitivities of the Stribeck curve to the macroscopic pin geometry and
the microscopic surface roughness. Therefore, diﬀerent roughness represen-
tations on the measured macroscopic pin geometry as well as variations of
the macroscopic geometry are considered in the following.
The parameters used for the simulations are summarized in Table 1. In
order to have a stationary numerical problem, the disc is assumed to be
11
perfectly ﬂat. Since it is of a harder material than the pin, the disc is also
assumed to be rigid. Consequently, elastic deformations are only modelled
for the pin surface.
Parameter Value Unit Description
E206 ·109Pa Young’s modulus
Np1012number of pin discretization cells
Nr3712number of roughness discretization cells
pa101325 Pa ambient pressure
pcav 80000 Pa cavitation pressure
U0.01...5 m/s disc velocity
tol 105relative error tolerance
µl0.0030758 Pa ·s dynamic viscosity of uncavitated lubricant
ν0.321 Poisson’s ratio
σY1200 ·106Pa yield strength
Table 1: Numerical parameters.
4.1. Inﬂuence of the roughness on the Stribeck curve
In section 3.1, two approaches of considering the surface roughness in the
Reynolds equation were presented and will be compared in the following.
The ﬁrst approach is simply using the meltdown gap height in the Reynolds
equation. The second method consists of using the homogenized Reynolds
equation. While the homogenization method allows the consideration of aver-
aged roughness eﬀects, it also increases the computational costs compared to
just using the meltdown height gap in the Reynolds equation. The reason is a
less sparse system matrix because of the oﬀ-diagonal homogenization factors
in Matrix Aand the additional interpolation of the homogenization factors.
Based on the measured pin and roughness proﬁles shown in the Figures 2
and 3, the Stribeck curves are computed with both roughness methods and
the obtained results are displayed Figure 8. Note that both methods use the
roughness proﬁle for the computation of the contact mechanics as described
in section 3.2. Their diﬀerence is only in the consideration of roughness ef-
fects on the hydrodynamics and it can be seen that almost identical results
are obtained. However, the computation using the homogenized Reynolds
equation took about 3 times longer than the other one. Since roughness
12
eﬀects on the hydrodynamics are apparently negligible for the presently con-
sidered roughness proﬁle, only the more eﬃcient concept of the meltdown
gap height is used in the following.
Nonetheless, it should be noted that the roughness proﬁle is important
for determining the onset of the contact pressure contribution which in turn
strongly inﬂuences the transition from the purely EHL to the mixed lubrica-
tion regime. If the gap coordinate h0at any point ~x above the pin becomes
less than the value of the highest roughness asperity Rp, surface contact oc-
curs through the contribution of the contact mechanics. Thus, the transition
point of the purely EHL to the mixed lubrication regime can be deﬁned as
the critical disc velocity Ucat which the minimum gap coordinate above the
pin min (h0(~x)) is equal to Rp. This is visualized in Figure 9 with the results
of the meltdown gap height method.
Figure 8: Stribeck curves using melt-
down gap height and homogenization.
Uc
mixed
lubrication
purely
EHL
Figure 9: Minimum gap coordinate
above the tribometer pin compared to
Rpas a function of the disc velocity.
4.2. Inﬂuence of the pin shape on the Stribeck curve
To investigate the inﬂuence of a diﬀerent macroscopic gap height, the
pin proﬁle is approximated by two parabolas. One parabola is designed to
closely ﬁt the measured pin proﬁle in the center while the other one is chosen
(within a parameter study) such that it captures the experimentally deter-
mined transition point from EHL to mixed lubrication in the Stribeck curve.
The corresponding pin proﬁles and computed Stribeck curves are depicted in
13
Figures 10 and 11. The predicted friction coeﬃcient is in very good agree-
ment with the experimental data [8] in the boundary regime while there is a
diﬀerence by about one order magnitude in the EHL regime. This diﬀerence
might be related to the neglected temperature and pressure dependence of
the ﬂuid viscosity on the numerical side or the low signal-to-noise-ratio of
the experiment in this regime. While the numerical predictions almost coin-
cide in EHL regime, the transition points from the purely EHL to the mixed
lubrication regime and the corresponding friction coeﬃcients show a clear
dependence on the diﬀerent macroscopic pin geometries. This dependence is
further investigated in the following. Therefore, the parabola center height
lcis deﬁned as the diﬀerence of the parabola proﬁle height in the center of
the pin to the proﬁle height at the pin’s rim. The pin parabola with a center
height of lc= 2µm is taken as a reference proﬁle because it ﬁts the experi-
mental data closely. Further simulations are carried out with pin parabolas
of the center heights lc= 1.9µm, lc= 1.5µm and lc= 1µm. They correspond
to a relative decrease of 5%, 25% and 50% of the reference center height. The
resulting Stribeck curves are displayed in Figure 12. Afterwards, the change
in the friction coeﬃcient relative to the reference proﬁle is computed for each
velocity as displayed in Figure 13. It shows that a measurement deviation of
25% or 0.5µm in the characteristic length of the reference pin causes a max-
imum diﬀerence in the friction coeﬃcient of more than 80%. Therefore, the
observed diﬀerence between the experimentally determined Stribeck curve
and the model prediction for the measured pin geometry in the mixed lubri-
cation regime could be caused by the fact that the macroscopic pin geometry
diﬀers for a run-in pin.
14
lc= 2µm
Figure 10: Gap coordinates for diﬀerent
pin proﬁles.
Figure 11: Stribeck curves for diﬀerent
pin proﬁles and the experimentally de-
termined Stribeck curve of Braun et al.
[8].
Figure 12: Stribeck curves obtained
by the variation of the reference pin
parabola.
Figure 13: Relative diﬀerence in the
friction coeﬃcient compared to the ref-
erence pin parabola.
5. Conclusions
In order to investigate the sensitivity of a predicted Stribeck curve we
studied the macroscopic pin shape and its microscopic roughness in a simpli-
15
ﬁed digital twin of a pin-on-disc tribometer. The numerical model consists
of the homogenized Reynolds equation with mass-conserving cavitation, the
boundary element method and the contact mechanics model of Bowden and
Tabor [23, 24]. The model is capable of simulating the tribometer in EHL,
mixed and boundary lubrication conditions.
The experimental results of Braun et al. [8], complemented by informa-
tion about the macroscopic pin geometry and surface roughness, serve as
reference data for the numerically predicted Stribeck curve.
The main ﬁndings of the work can be summarized as follows:
The measured surface roughness that is employed in this work has neg-
ligible inﬂuence on the tribometer’s Stribeck curve in the EHL regime
but signiﬁcantly aﬀects the relative velocity at which the transition
from the purely EHL to the mixed lubrication regime occurs.
The mixed lubrication regime of the Stribeck curve is highly sensitive
to the macroscopic pin geometry. An approximation of the pin proﬁle
through a parabola allows a quantitative estimation of the pin geometry
inﬂuence on the Stribeck curve. It indicates that a 0.5µm variation in
the characteristic length of the pin proﬁle can cause a deviation in the
predicted friction coeﬃcient of more than 80%.
In order to enable the comparison of experimental results with numeri-
cal predictions, which is critical for the further development of a digital
twin of a pin-on-disc-tribometer, macro- and microscopic surface mea-
surements of the pin and disc proﬁles in worn conditions need to be
provided.
We note that further steps to create a digital twin do not only require the
knowledge of the geometric properties discussed in this work in addition to
the values listed in Table 1. Moreover, the self aligning pin holder may allow
variations of the pin inclination which induces changes in the gap height
distribution. Also geometrical imperfections and EHL eﬀects on the disc
should be considered.
Acknowledgments
The authors want to express their gratitude to Dr. Paul Schreiber for
providing the measurements of the representative roughness used by Braun
16
et al. [8]. This document is the result of a research project funded by the
Friedrich und Elisabeth Boysen-Foundation.
17
Appendix A. Homogenization
The homogenization factors A,~
b,~c and dare determined as [15, 22]:
A(h0) =
1
Y1Y2R
A~
ξ
h31 + ∂χ1
∂ξ1dA~
ξ
1
Y1Y2R
A~
ξ
h3∂χ2
∂ξ1dA~
ξ
1
Y1Y2R
A~
ξ
h3∂χ1
∂ξ2dA~
ξ
1
Y1Y2R
A~
ξ
h31 + ∂χ2
∂ξ2dA~
ξ
, (A.1)
~
b(h0) =
1
Y1Y2R
A~
ξ
h3∂χ3
∂ξ16lUdA~
ξ
1
Y1Y2R
A~
ξ
h3∂χ3
∂ξ2dA~
ξ
, (A.2)
~c(h0) =
1
Y1Y2R
A~
ξh
21 + ∂χ1
∂ξ1dA~
ξ
1
Y1Y2R
A~
ξh
2
∂χ2
∂ξ1dA~
ξ
, (A.3)
d(h0) = 1
Y1Y2Z
A~
ξ
h
2
∂χ3
∂ξ1
+µlU
hdA~
ξ. (A.4)
Bayada et al. [15] only employed longitudinal, transversal or oblique
roughness proﬁles and could therefore derive an analytical solution of the
homogenization factors. In order to consider arbitrary roughness proﬁles,
the follwoing local problems must be solved numerically for their solutions
χ1,χ2and χ3in [22]:
~
ξ·h3~
ξχ1=∂h3
∂ξ1
, (A.5)
~
ξ·h3~
ξχ2=∂h3
∂ξ2
, (A.6)
~
ξ·h3~
ξχ3= 6µlU·∂h
∂ξ1
. (A.7)
The indexed nabla operator ~
ξmeans using it with respect to the ~
ξ
coordinates.
Appendix B. Relation of Vickers hardness and yield strength
Depending on whether the Vickers hardness is calculated from the in-
dentation mass or force, there are two slightly diﬀerent hardness deﬁnitions.
18
Tabor describes the Vickers hardness HVTas [23]:
HVT=W
AS
= 0.9272 W
AG
= 1.8544W
d2, (B.1)
where Wis the applied indentation mass in kg and ASis the area of
the sides of the indented volume. For a Vickers indentation pyramid, AS=
AG/0.927 holds, where AG=d2/2 is the projected surface of the indented
volume. The diagonal length of AGis given by din mm. The resulting hard-
ness HVThas the units kg/mm2. In this case, the ultimate tensile strength
in MPa roughly correlates to:
U T S 3HVT·m/s2. (B.2)
However, there exists another deﬁnition of the Vickers hardness HV in
dependency of the indentation foce FIin the unit N, which is commonly used
e.g. in Germany [26]:
HV = 0.189FI
d2. (B.3)
From experiments, an approximation for the ultimte tensile strength in
MPa was found [26]:
U T S 3HV . (B.4)
In both cases, the yield strength is equal to the ultimate tensile strength
σy=U T S if ideal elastic-plastic material properties are assumed.
19
[1] K. Holmberg, A. Erdemir, The impact of tribology on energy use and
CO2 emission globally and in combustion engine and electric cars, Tri-
bology International 135 (2019) 389–396.
[2] D. Gropper, L. Wang, T. J. Harvey, Hydrodynamic lubrication of tex-
tured surfaces: A review of modeling techniques and key ﬁndings, Tri-
bology International 94 (2016) 509–529.
[3] I. Etsion, State of the art in laser surface texturing, Journal of Tribology
127 (1) (2005) 248–253.
[4] M. B. Dobrica, M. Fillon, About the validity of Reynolds equation and
inertia eﬀects in textured sliders of inﬁnite width, Proceedings of the
Institution of Mechanical Engineers, Part J: Journal of Engineering Tri-
bology 223 (1) (2009) 69–78.
[5] A. Codrignani, B. Frohnapfel, F. Magagnato, P. Schreiber, J. Schneider,
P. Gumbsch, Numerical and experimental investigation of texture shape
and position in the macroscopic contact, Tribology International 122
(2018) 46–57.
[6] A. Dadouche, M. Fillon, W. Dmochowski, Performance of a hydrody-
namic ﬁxed geometry thrust bearing: comparison between experimental
data and numerical results, Tribology Transactions 49 (3) (2006) 419–
426.
[7] F. Proﬁto, S.-C. Vladescu, T. Reddyhoﬀ, D. Dini, Experimental valida-
tion of a mixed-lubrication regime model for textured piston-ring-liner
contacts, Materials Performance and Characterization 6 (2) (2017) 112–
129.
[8] D. Braun, C. Greiner, J. Schneider, P. Gumbsch, Eﬃciency of laser
surface texturing in the reduction of friction under mixed lubrication,
Tribology International 77 (2014) 142–147.
[9] J. Schneider, D. Braun, C. Greiner, Laser textured surfaces for mixed
lubrication: inﬂuence of aspect ratio, textured area and dimple arrange-
ment, Lubricants 5 (3) (2017) 32.
[10] D. Z. Segu, P. Hwang, Friction control by multi-shape textured surface
under pin-on-disc test, Tribology International 91 (2015) 111–117.
20
[11] B. Podgornik, L. Vilhena, M. Sedlaˇcek, Z. Rek, I. ˇ
Zun, Eﬀectiveness and
design of surface texturing for diﬀerent lubrication regimes, Meccanica
47 (7) (2012) 1613–1622.
[12] H. Zhang, D. Zhang, M. Hua, G. Dong, K. Chin, A study on the tri-
bological behavior of surface texturing on babbitt alloy under mixed or
starved lubrication, Tribology Letters 56 (2) (2014) 305–315.
[13] A. Kovalchenko, O. Ajayi, A. Erdemir, G. Fenske, I. Etsion, The eﬀect
of laser surface texturing on transitions in lubrication regimes during
unidirectional sliding contact, Tribology International 38 (3) (2005) 219–
225.
[14] R. Larsson, Modelling the eﬀect of surface roughness on lubrication in
all regimes, Tribology International 42 (4) (2009) 512–516.
[15] G. Bayada, S. Martin, C. V´azquez, An average ﬂow model of the
Reynolds roughness including a mass-ﬂow preserving cavitation model,
Journal of Tribology 127 (4) (2005) 793.
[16] T. Woloszynski, P. Podsiadlo, G. W. Stachowiak, Eﬃcient solution to
the cavitation problem in hydrodynamic lubrication, Tribology Letters
58 (1) (2015) 839.
[17] R. Pohrt, Q. Li, Complete boundary element formulation for normal
and tangential contact problems, Physical Mesomechanics 17 (4) (2014)
334–340.
[18] C. Greiner, T. Merz, D. Braun, A. Codrignani, F. Magagnato, Optimum
dimple diameter for friction reduction with laser surface texturing: the
eﬀect of velocity gradient, Surface Topography: Metrology and Proper-
ties 3 (4) (2015) 044001.
[19] B. Bhushan, Surface roughness analysis and measurement techniques,
in: Modern Tribology Handbook, Two Volume Set, CRC press, 2000,
pp. 79–150.
[20] C. Shen, M. Khonsari, On the magnitude of cavitation pressure of
steady-state lubrication, Tribology Letters 51 (1) (2013) 153–160.
21
[21] P. Forooghi, A. Stroh, P. Schlatter, B. Frohnapfel, Direct numerical sim-
ulation of ﬂow over dissimilar, randomly distributed roughness elements:
A systematic study on the eﬀect of surface morphology on turbulence,
Physical Review Fluids 3 (4) (2018) 044605.
[22] F. Sahlin, R. Larsson, A. Almqvist, P. Lugt, P. Marklund, A mixed
lubrication model incorporating measured surface topography. Part 1:
theory of ﬂow factors, Proceedings of the Institution of Mechanical Engi-
neers, Part J: Journal of Engineering Tribology 224 (4) (2009) 335–351.
[23] D. Tabor, The Hardness of Metals, At the Clarendon Press, Oxford,
1951.
[24] F. P. Bowden, D. Tabor, The Friction and Lubrication of Solids, 2nd
Edition, At the Clarendon Press, Oxford, 1954.
[25] A. E. H. Love, The stress produced in a semi-inﬁnite solid by pressure
on part of the boundary, Philosophical Transactions of the Royal Society
of London. Series A, Containing Papers of a Mathematical or Physical
Character 228 (659-669) (1929) 377–420.
[26] W. Weissbach, M. Dahms, C. Jaroschek, Werkstoﬀe und ihre Anwen-
dungen : Metalle, Kunststoﬀe und mehr, Springer Vieweg, Wiesbaden,
2018.
22
... Attention is paid to double the size of the kernel in each direction and to zero pad the hydrodynamic pressure field such that a linear instead of a circular convolution is obtained. After the convolution, the deformation and pressure fields are resized to their original size [23,32,33]. After computing ⃗ G and ⃗ F , the Newton-Raphson method is used to determine the updates of non-dimensional relative pressure ⃗ p * and cavity fraction ⃗ [10]: ...
Article
Full-text available
Many engineering applications rely on lubricated gaps where the hydrodynamic pressure distribution is influenced by cavitation phenomena and elastic deformations. To obtain details about the conditions within the lubricated gap, solvers are required that can model cavitation and elastic deformation effects efficiently when a large amount of discretization cells is employed. The presented unsteady EHL-FBNS solver can compute the solution of such large problems that require the consideration of both mass-conserving cavitation and elastic deformation. The execution time of the presented algorithm scales almost with Nlog(N)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$N\log (N)$$\end{document} where N is the number of computational grid points. A detailed description of the algorithm and the discretized equations is presented. The MATLAB©\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{\copyright }$$\end{document} code is provided in the supplements along with a maintained version on GitHub to encourage its usage and further development. The output of the solver is compared to and validated with analytical, simulated, and experimental results from the literature to provide a detailed comparison of different discretization schemes of the Couette term in presence of gap height discontinuities. As a final result, the most favorable scheme is identified for the unsteady study of surface textures in ball-on-disc tribometers under EHL conditions.
... Attention is paid to double the size of the kernel in each direction and to zero pad the hydrodynamic pressure field such that a linear instead of a circular convolution is obtained. After the convolution, the deformation and pressure fields are resized to their original size [23,31,32]. ...
Preprint
Full-text available
Many engineering applications rely on lubricated gaps where the hydrodynamic pressure distribution is influenced by cavitation phenomena and elastic deformations. To obtain details about the conditions within the lubricated gap, solvers are required that can model cavitation and elastic deformation effects efficiently when a large amount of discretization cells is employed. The presented unsteady EHL-FBNS solver can compute the solution of such large problems that require the consideration of both mass-conserving cavitation and elastic deformation. The execution time of the presented algorithm scales almost with N log( N ) where N is the number of computational grid points. A detailed description of the algorithm and the discretized equations is presented. The MATLAB © code is provided in the supplements along with a maintained version on GitHub to encourage its usage and further development. The output of the solver is compared to and validated with simulated and experimental results from the literature to provide a detailed comparison of different discretization schemes of the Couette term in presence of gap height discontinuities. As a final result, the most favourable scheme is identified for the unsteady study of surface textures in ball-on-disc tribometers under severe EHL conditions.
... Empirically, it has been found that, in the hydrodynamic regime, the friction coefficient μ (the ratio of friction force and normal force) scales with the "conventional" Hersey number, defined as Hr ≡ ðηU 0 =F N Þ, where η, U 0 , and F N are the viscosity, sliding velocity, and normal force, respectively. The scaling has been suggested to be linear [5,11], although more recent work uses a power-law relation [12][13][14][15]. ...
Article
Most frictional contacts are lubricated in some way, but is has proven difficult to measure and predict lubrication layer thicknesses and assess how they influence friction at the same time. Here we study the problem of rigid-isoviscous lubrication between a plate and a sphere, both experimentally and theoretically. The liquid layer thickness is measured by a novel method using inductive sensing, while the friction is measured simultaneously. The measured values of the layer thickness and friction on the disk are well described by the hydrodynamic description of liquid flowing through a contact area. This allows us to propose a modified version of the Hersey number that compares viscous to normal forces and allows us to rescale data for different geometries and systems. The modification overcomes the shortcomings of the commonly used Hersey number, adds the effects of the geometry of the configuration on the friction, and successfully predicts the lubrication layer thickness.
Thesis
Full-text available
The growing demand for higher performance and lower environmental impact of tribological devices has led to the development of textured surfaces as efficient way to reduce friction losses and wear. Great part of the tribological research of such engineered surfaces relies on experimental set-ups based on pin-on-disc tribometers. The aim of the present work is the numerical representation of such kind of experimental set-ups in order to study the influence of textures in the hydrodynamic regime. In the initial analysis of the mentioned set-up, particular attention is paid to the high-resolution representation of the macroscopic geometry of the contact. In this regard, the influence of the velocity gradient in the plane of rotation is investigated and compared to experimental studies. The role of inertia effects is then examined by means of the Navier-Stokes equation and the validity of the Reynolds equation is investigated for 3D dimples resulting in the extension of previous literature results. The effects of viscous dissipation are also considered in order to assess their influence on the pressure losses in the contact, which are found to scale with the square of the Sommerfeld number. Finally, the influence of the operating conditions on the optimal shape parameter of a texture made by dimples is investigated by means of an extensive parametric study. As a result, the dimple depth is shown to have a higher impact on the texture performance than the diameter. Moreover, the dimple depth, which corresponds to the minimal friction coefficient, is found to scale with the square root of the Sommerfeld number in agreement with experimental investigations on the same tribometer.
Article
Full-text available
In this work, the influence of operating conditions on the shape parameters of surface texture is investigated by means of both numerical and experimental investigations. The analysed texture consists of micro-dimples obtained through laser surface texturing on a pin-on-disc configuration. From the numerical point of view, particular attention is paid to the faithful representation of the 2D surface of the experimental set-up and to modelling cavitation phenomena through a mass conserving algorithm. As results, the dimple depth shows a higher relevance than diameter in determining the optimal texture shape (both in terms of friction reduction and load carrying capacity). Moreover, the dimple depth, corresponding to the minimal friction, is found to scale with the square root of the Sommerfeld number in agreement with the experimental results. Finally, it is found that a numerical approach with the present hydrodynamic model cannot account for friction reduction obtained experimentally with different orientation of the texture.
Article
Full-text available
Unidirectional sliding experiments with polished and laser textured steel surfaces were carried out to investigate the effects of different textured area densities, aspect ratios and dimple arrangements. The system was lubricated with Polyalphaolefin (PAO) at 100 °C and the contact pressure was 3 MPa. For measuring Stribeck curves, the sliding speed was controlled between 40 and 2000 mm/s. The textured area density was varied between 5% and 30%, with the lowest friction values found for 10%. Aspect ratios ranging from 0.02 to 0.2 were investigated and for 0.1 the lowest friction values were measured. The dimple arrangements tested were cubic, hexagonal and a random distribution for a textured area density of 10% and an aspect ratio of 0.1. Our results demonstrate that the dimple arrangement does affect friction values, hinting to the fact that individual texture elements do influence each other. The optimum dimple arrangement was found in a hexagonal packing. This systematic variation of these three key texturing parameters for the morphological texturing of a tribological surface with dimples will allow a strategic choice of texturing parameters. This makes the most of the tremendous potential that laser surface texturing has for reducing friction forces and thereby CO2 emissions.
Article
Full-text available
Abstract Understanding the influence of surface properties (roughness, grooves, discrete textures/dimples) on the performance of hydrodynamically lubricated contacts has been the aim of numerous studies. A variety of different numerical models have been employed by many researchers in order to find optimal texturing parameters (shape, size, distribution) for best performance enhancement in terms of load carrying capacity, film thickness, friction and wear. However, the large number of different modeling techniques and complexity in the patterns make finding the optimum texture a challenging task and have led to contrary conclusions. This article outlines the research effort on surface texturing worldwide, reviews the key findings and, in particular, provides a comparative summary of different modeling techniques for fluid flow, cavitation and micro-hydrodynamic effects.
Article
Full-text available
The morphological texturing of surfaces has demonstrated high potential to reduce friction and wear. In order to understand the effect of different velocity gradients over the textured area on the optimum dimple diameter, we textured brass pins with round dimples having diameters between 20 and 200 μm. The dimple depth and packing density were kept constant. The samples were tested in a pin-on-disc fashion against sapphire discs and experiments were conducted under mixed lubrication and for two different sliding radii. Our results show that larger velocity gradients favor smaller dimples, whereas for the smaller velocity gradients, larger dimple diameters were beneficial. The effect of there being an influence of the velocity gradient was also found in computational fluid dynamics (CFD) simulations. Experimentally, friction forces could be reduced by up to 80%, demonstrating the tremendous potential of laser surface texturing (LST) to lower friction forces and reduce CO 2 emissions.
Article
Growing concerns over energy and environmental sustainability have lately sparked worldwide interest in more efficient and cleaner transportation systems and industrial activities. Friction roughly consumes one-fifth of all energy used worldwide. One-third of all energy used in transportation goes to overcome friction. At the same time, the fruits of decades of dedicated research on all-electric vehicles powered by advanced batteries are paving the way toward a much cleaner and sustainable transportation future. In this article, we provide a short overview of what are the energy efficiency and environmental impacts of current transportation, industrial, and residential systems and how much of that efficiency is adversely affected by friction and wear losses in moving mechanical parts and components. We also touch upon recent advances in new materials, lubricants, and design changes that could reduce energy losses by 18–40%, mainly resulting from friction and wear. The savings would be up to 8.7% of the total global energy use and 1.4% of the gross national products (GNP). Finally, we calculate the energy consumption and friction losses in battery-powered electric passenger cars and show the benefit of electric cars where the total energy use is in average 3.4 times lower compared to combustion engine powered cars. The CO 2 emissions are 4.5 times higher for a combustion engine car compared to an electric car when the electricity comes from renewable energy sources. Moving from fossil to renewable energy sources may cut down the energy losses due to friction in energy production by more than 60%.
Article
Direct numerical simulations are used to investigate turbulent flow in rough channels, in which topographical parameters of the rough wall are systematically varied at a fixed friction Reynolds number of 500, based on a mean channel half-height h and friction velocity. The utilized roughness generation approach allows independent variation of moments of the surface height probability distribution function [thus root-mean-square (rms) surface height, skewness, and kurtosis], surface mean slope, and standard deviation of the roughness peak sizes. Particular attention is paid to the effect of the parameter Δ defined as the normalized height difference between the highest and lowest roughness peaks. This parameter is used to understand the trends of the investigated flow variables with departure from the idealized case where all roughness elements have the same height (Δ=0). All calculations are done in the fully rough regime and for surfaces with high slope (effective slope equal to 0.6–0.9). The rms roughness height is fixed for all cases at 0.045h and the skewness and kurtosis of the surface height probability density function vary in the ranges −0.33 to 0.67 and 1.9 to 2.6, respectively. The goal of the paper is twofold: first, to investigate the possible effect of topographical parameters on the mean turbulent flow, Reynolds, and dispersive stresses particularly in the vicinity of the roughness crest, and second, to investigate the possibility of using the wall-normal turbulence intensity as a physical parameter for parametrization of the flow. Such a possibility, already suggested for regular roughness in the literature, is here extended to irregular roughness.
Article
Recent experiments have shown that automotive piston-liner friction may be reduced by up to 50 % if the surface of the liner is laser textured with certain configurations of micro-pockets. It is important to model this behavior to understand and optimize the friction reduction mechanisms that are occurring. However, until now, very few models that predict the lubrication performance of textured surfaces have been successfully validated against experimental data. This is because of the requirement for them to: (1) reproduce experimental configurations with a certain degree of fidelity, (2) conserve mass properly, and (3) account for transient, boundary lubrication conditions. To address this, the current paper presents a comparison between the results from a numerical model, which fulfils these criteria, and an experimental test rig operating under the same conditions. The mathematical modeling is based on the averaged Reynolds' equation with Patir and Cheng's flow factors and the p - θ Elrod-Adams mass-conserving cavitation model. Simultaneously to the fluid flow solution, the contact pressures that arise from the asperity interactions are also included into the calculations through the well-known stochastic Greenwood and Tripp model for rough contacts. The experimental data is produced using a reciprocating tribometer, whose contact conditions are closely controlled and accurately mimic those found in an automotive piston-liner conjunction. Data is presented in terms of friction force versus stroke angle, and the similarities and differences between the model and experiment are discussed.