ISSN 1220-8434 Volume 18, 2010

ACTA TRIBOLOGICA
Volume 18, 2010
Acta tribologica
A Journal on the Science of Contact Mechanics, Friction, Lubrication,
Wear, Micro/Nano Tribology, and Biotribology
Volume 18, 2010
EDITOR E. Diaconescu, University of Suceava, ROMANIA
EDITORIAL BOARD N.N. Antonescu, Petroleum-Gas University of Ploiesti, ROMANIA
J.R. Barber, University of Michigan, U.S.A
Y. Berthier, INSA de Lyon, FRANCE
M. Ciavarella, Politecnico di Bari, ITALY
T. Cicone, University Politehnica of Bucharest, ROMANIA
S. Cretu, Technical University of Iasi, ROMANIA
L. Deleanu, University of Galati, ROMANIA
D. Dini, Imperial College London, UNITED KINGDOM
V. Dulgheru, Technical University of Moldova, MOLDOVA
I. Etsion, Technion, Haifa, ISRAEL
M. Glovnea, University of Suceava, ROMANIA
R. Glovnea, University of Sussex, UNITED KINGDOM
I. Green, Georgia Institute of Technology, U.S.A
M. Khonsari, Louisiana State University, U.S.A
Y. Kligerman, Technion, Haifa, ISRAEL
D. Nelias, INSA de Lyon, FRANCE
D. Olaru, Technical University of Iasi, ROMANIA
M. Pascovici, University Politehnica of Bucharest, ROMANIA
M. Ripa, University of Galati, ROMANIA
A. Tudor, University Politehnica of Bucharest, ROMANIA
ASSISTANT EDITOR S. Spinu, University of Suceava, ROMANIA
Published by the Applied Mechanics Section of the University of Suceava
University Stefan cel Mare of Suceava Publishing House
13
th
University Street, Suceava, 720229, Suceava, ROMANIA
Phone: (40) 0230 216 147 int. 273, E-mail: editura@usv.ro
ACTA TRIBOLOGICA VOLUME 18, 2010
CONTENTS
1 A. URZIC, S. CRETU
A Numerical Procedure to Generate Non-Gaussian Rough Surfaces
7 C. CIORNEI, E. DIACONESCU
Preliminary Theoretical Solution for Electric Contact Resistance between
Rough Surfaces
12 C.-I. BARBINTA, S. CRETU
The Influence of the Rail Inclination and Lateral Shift on Pressure Distribution
in Wheel - Rail Contact
19 C. SUCIU, E. DIACONESCU
Preliminary Theoretical Results upon Contact Pressure Assessment by Aid of
Reflectivity
27 S. SPINU
Numerical Simulation of Elastic-Plastic Contact
34 Y. NAGATA, R. GLOVNEA
Dielectric Properties of Grease Lubricants
42 J. PADGURSKAS, R. KREIVAITIS, A. KUPINSKAS, R. RUKUIA,
V. JANKAUSKAS, I. PROSYEVAS
Influence of Nanoparticles on Lubricity of Base Mineral Oil
46 A.V. RADULESCU, I. RADULESCU
Influence of the Rheometer Geometry on the Rheological Properties of
Industrial Lubricants
52 V.-F. ZEGREAN, E. DIACONESCU
Measurement of Lubricant Oil Microviscosity Based on Resonant Frequency
Shift of AFM Cantilever
58 M.C. CORNECI, A.-M. TRUNFIO-SFARGHIU, F. DEKKICHE, Y.
BERTHIER, M.-H. MEURISSE, J.-P. RIEU
Influence of Lubricant Physicochemical Properties on the Tribological
Operation of Fluid Phase Phospholipid Biomimetic Surfaces
65 S. LE FLOCH, M.C. CORNECI, A.-M. TRUNFIO-SFARGHIU, M.-H.
MEURISSE, J.-P. RIEU, J. DUHAMEL, C. DAYOT, F. DANG, M. BOUVIER,
C. GODEAU, A. SAULOT, Y. BERTHIER
Imagerie Medicale pour Evaluer les Conditions du Fonctionnement
Tribologiques des Articulations Synoviales
77 M.C. CORNECI, A.-M. TRUNFIO-SFARGHIU, F. DEKKICHE,
Y. BERTHIER, M.-H. MEURISSE, J.-P. RIEU, M. LAGARDE,
M. GUICHARDANT
Phospholipides dans le Fluid Synovial - Influence sur le Fonctionnement
Tribologique des Articulations Synoviales Pathologiques
85 I.C. ROMANU, E. DIACONESCU
Bioarticular Friction
89 A.-M. TRUNFIO-SFARGHIU, M.C. CORNECI, Y. BERTHIER,
M.-H. MEURISSE, J.-P. RIEU
Mechanical and Physicochemical Analysis of the Tribological Operation of
Joint Replacements
106 D. N. OLARU, C. STAMATE, A. DUMITRASCU, G. PRISACARU
Rolling Friction Torque in Microsystems
113 L. DELEANU, S. CIORTAN
Evaluating Tribological Damages by 3D Profilometry
120 M. RP, S. BOICIUC
Characterisation of Laser Cladding with NiCrBFeAl Alloy by
Profilometric Study of the Scratch Tracks
128 M. VLASE, A. TUDOR
An Analytical Wear Model of the Pipes for Concrete Transportation
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 1-6





Ana URZIC
e-mail: sf_ana8107@yahoo.com

Spiridon CREU
e-mail: spcretu@mail.tuiasi.ro


Department of Machine Design,
Technical University of Iai,
ROMANIA

A NUMERICAL PROCEDURE TO GENERATE
NON-GAUSSIAN ROUGH SURFACES

The paper presents an algorithm for computer simulation of non-
Gaussian surfaces. By using a random number generator, a input
matrix is formed as a first representation of a Gaussian roughness
with zero mean, and unit standard deviation. The autocorrelation
function was assumed to have an exponential form. To fulfill this
requirement, in the first step, the matrix containing the roughness
heights was obtained by a linear transformation of the input matrix.
In the second step the skewness and kurtosis of the input sequence
have been established for the desired skewness and kurtosis of an
output sequence. Finally the non-Gaussian random series have been
generated by using the Johnson translator system.
The numerical results pointed out that the developed algorithm can
be further used to simulate manufacturing processes that produce
real surfaces which may present a non-Gaussian distribution, as
well as the abrasive wear and running in phenomena.
Keywords: roughness, autocorrelation, skewness, kurtosis


1. INTRODUCTION

Both experimental and numerical studies
have pointed out that roughness acts as stress
concentration sites and induce stresses greater than
in an equivalent smooth contact. The real areas of
contact and the asperity contact pressures are
essential parameters for any wear modeling. These
parameters can vary significantly depending on
surface topography. A small change in the
distribution of heights, wave length and curvature of
the surface roughness can have a noticeable effect
on the deformation behaviors of the rough surfaces.
Manufacturing processes produce real
surfaces which are sometimes quite different from
Gaussian distribution. For example, a lathe turned
surface is far from random; its peaks are nearly all
the same height and its valley nearly all the same
depth. A ground surface which is subsequently
polished so that the tips of the higher asperities are
removed departs markedly from being Gaussian.

Figure 1. The changes in profile caused by running
in and abrasive wear
Similar profiles are presented by surfaces that
had carried out abrasive wear or running in
processes, Figure 1.
Any parametric study involving roughness
requires surfaces with known statistical proprieties
and it is much more convenient to generate them
numerically rather to measure manufactured rough
surfaces. An essential requirement for any
numerical algorithms for roughness simulation is
their abilities to generate rough surface which have
statistical proprieties similar to real surfaces.
Most of the statistical proprieties of a rough
surface can be derived from knowledge of two
statistical functions: the frequency density function
and the autocorrelation function, Bakolas V. [1],
Bushan B. [2], Greenwood J.A. [6], Robbe -
Valloire F. [11]. J.Mc.Cool [9] shows that it is
possible to describe any statistical distribution
through knowledge of only four central moments of
cumulative distribution function of probabilities.
Consequently, a good algorithm should be
able to generate surfaces having prescribed
frequency density functions and autocorrelation
functions.
The developed procedure starts from the
imposed values for the normalized central moments:
the mean height,
a
R , standard deviation
q
R ,
skewness parameter Sk, kurtosis parameter K, as
well as for the correlation lengths x, y, of the
autocorrelation function.

2
2. SPATIAL AND SPECTRAL PARAMETERS
OF ROUGHNESS

2.1 Probability Density Function (PDF)
If for convenience z was measured from the
mean plane of the surface, then the height z(x, y) of
a rough surface may be considered as a two-
dimensional random variable. The spatial
characteristics can be adequately described with the
use of probability function p(z) which denotes the
probability that a point on the surface has a height
equal to z. It has been found, Bakolas V. [1],
Bhushan B. [2], Patir N. [10], that many real
surfaces, notably freshly grounded surfaces, reveal a
height distribution which is close to the normal
Gaussian probability function:

2
2
z
p(z) exp
2 2
| |

=
|
\ .
, (1)

where is the standard (r.m.s.) deviation from the
mean height.
The shape of the probability function can
provide useful information about the nature of the
roughness profile. A mathematical presentation of
this shape is provided by the moments of the
probability density function about the mean.

2.2 The normalized moments of PDF
The first normalized central moment is the
mean height,
a
R which is generally removed before
data processing and is therefore zero:

a
R z p(z) dz

. (2)

The second moment is the variance
2
q
R of
the roughness heights, meaning the standard
deviation
q
R , or the root mean square (r.m.s.) , of
the surface heights:

2 2
z p(z) dz

. (3)


The third normalized central moment is
called skewness:

3
3
1
Sk z p(z) dz

. (4)

The skewness parameter represents a measure
of the symmetry of the statistical distribution.
Symmetrical distributions have skewness equal to 0,
which means that they have evenly distributed peaks
and valleys of specific height. Profiles with high
peaks and shallow valleys present a positive
skewness, while profiles with larger valleys than
peaks present a negative skewness, Figure 2.


Figure 2. Profiles with different degrees of
asymmetry and the shapes of PDF


The fourth normalized central moment is
called kurtosis:

4
4
1
K z p(z) dz

. (5)

Kurtosis represents the spikiness of the
statistical distribution and is a measure of the degree
of pointedness or bluntness, Figure 3. Symmetric
Gauss distribution has a kurtosis of 3, Figure 4.



Figure 3. Profiles with different kurtosis values and
the shapes of PDF



Figure 4. PDF for random distributions with
different skewness values (a),
and for symmetrical distributions (Sk=0) with
different kurtosis values (b)

Typical skewness and kurtosis envelopes for
various manufacturing technologies are presented in
the Figure 5.

3



Figure 5. Skewness and kurtosis values for some
manufacturing technologies


3. NUMERICAL GENERATION OF NON-
GAUSSIAN RANDOM SURFACES

A 2D digital filter, as suggested by Hu and
Tonder [8], has been involved to change the input
sequence (k, )

into an output sequence z(I, J) :

n 1 m 1
k 0 0
z(I, J) h(k, ) (I k, J )

= =
=

,
(6)

I 0,1,..., N 1; = J 0,1,..., M 1; = n N/ 2 =
m M/ 2 = ,

where the h(k, ) is the digital filter function.
To establish the filter function h(k, ) the
following steps has to be fulfilled:
1. Obtain the autocorrelation function (ACF) and
the power spectral density (PSD) for the input
sequence .
2. Simulate a random matrix (surface) with a
negative exponential function for the ACF.
3. Obtain the PSD for the new random matrix with
negative exponential function for the ACF.
4. Determine the digital filter function.
5. Determine the needed skewness Sk

and
kurtosis K

of the input sequence for desired

skewness and kurtosis of an output sequence.
6. Generate a non-Gaussian random series
'
by
using Johnson translator system.
The steps 1 4 have been previously
presented by Cretu [3,4] and will not be further
discussed in the present paper.

3.1 Determine the skewness Sk

and kurtosis
K

of the input sequence

To generate a rough surface which has a non-
Gaussian distribution, the procedure will be to
transform the input sequence into a output sequence
which has similar values with the values we wish to
impose for skewness and kurtosis. To obtain the
skewness,
z
Sk and kurtosis,
z
K close to the required
values, the following equation was used:

q
3
i
i 1
z
3
q
2
2
i
i 1
Sk Sk
=

=
| |

|
\ .

; (7)
q q 1 q
4 2 2
i i j
i 0 i 0 j i 1
z
2
q
2
i
i 1
K 6
K

= = = +
=
+
=
| |

|
\ .

. (8)

These relationships have been proposed by
Watson and Spendding [12] and are valid for linear
transformation of the form:

z 0 x 1 x 1 2 x 2
q 1 x q 1 q x q
x
... ;

+
= + + +
+ +
(9)

where
z
Sk ,
z
K and are the required skewness and
kurtosis, and Sk

, K

are the input skewness and

kurtosis for Johnsons translator system, and

i
h(k, ); = (10)
i (k 1)m = + ; k 1, 2,..., n = ; 1, 2,..., m =

The validity of the following algebraic
equation:

( )
2
q 1 q q q
2 2 2 4
i j i i
i 1 j i 1 i 1 i 0
1
2

= = + = =
(
| |
(
=
|
(
\ .


(11)

allows to obtain a simpler form for the kurtosis
parameter:

( )
q
4
i
i 0
z
2
q
2
i
i 1

K K 3 3

=
= +
| |
|
\ .

. (12)

4
When arbitrary skewness and kurtosis are set, they
must fulfill the following relationship:

z z
K Sk 1 0 . (13)

3.2 Generation of the non-Gaussian random
series
'
by using the Johnson translator system
The non-Gaussian random series with
different skewness can be generated by using the
Johnson translator system. The Johnson system was
presented in their works by W.P.Elderton and
N.L.Johnson [5] and V.Bakolas [1]. The Johnson
system of frequency curves is based on the method
of moments and provides some curves that can be
used to generate a random distribution for which the
four moments are know. The Johnson system uses
three main conversion curves: S
U
, S
B
and S
L
:

U
S :
'
sinh

= +
| |

|

\ .
; (14)
L
S :
'
ln
| |

= +
|

\ .

( )
'
> ; (15)
B
S :
'
'
ln
| |

= +
|
+
\ .
( )
'
< < + (16)

where:
is a sequence of random numbers with
normal distribution, m 0 = , 1 = ,
Sk 0 = and K 3 = ;

'
is the sequence of random number derived
with desired values for the parameters
skewness and kurtosis, Sk

and K

;
, , and are constants to be determined
for the first four given moments by using
method of moments.
The initial distribution of random number had
to be chosen to follow a statistical distribution that
ensures the following constraints:
the average value is zero, ( ) m 0 = ;
standard deviation equal to unity, ( ) 1 = ;
the required value for skewness parameter,
Sk;
the required value for kurtosis parameter, K.


4. RESULTS

Three-dimensional surfaces maps of the non-
Gaussian random numbers with zero mean and unit
variance and the autocorrelation length x = y = 1
m, but different skewness (Sk) and kurtosis (K)
values, are presented successively in the Figure 6.

a.


b.
Figure 6. 3D random matrices with different
a. skewness (Sk); b. kurtosis (K)

In Figure 6a the skewness parameter was
changed between limits while the kurtosis function
maintained the value K 3 = . In the same manner, in
Figure 6b the kurtosis parameter was changed
between limits while the skewness function
maintained the value Sk 0 = .

5

a. b.
Figure 7. 2D profiles of the random matrices with different (a) skewness (Sk) and (b) kurtosis (K)

Figure 8. 3D random simulation for a worn surface and the corresponding 2D profiles

To highlight the effect of varying the
skewness and kurtosis parameters on the general
shape of the profile, the Figure 7 presents extracted
profiles along the x-x direction of surfaces
represented in Figure 6.
A worn surface is characterized by negative
values for the skewness function while the kurtosis
function has values equal or greater than 3, so that
the values 1 = Sk and 3 = K have been chosen for
the numerical simulation.
The Gaussian random matrix is presented in
the Figure 8a, while the non-Gaussian random
matrix with imposed values for the skewness and
kurtosis functions is presented in the Figure 8b. The
correspondent profiles of the two matrices are given
in the Figure 8c.


5. CONCLUSIONS

1. Manufacturing processes provide real
surfaces that may be quite different from Gaussian.
A ground surface which is subsequently polished
departs markedly from being Gaussian; similar

6
surfaces are caused by abrasive wear or running in
processes.
2. By using a random number generator an
input matrix is formed as a first representation of a
Gaussian roughness with zero mean, ( ) 0 = m , and
unit standard deviation, ( ) 1 = . The autocorrelation
function was assumed to have an exponential form.
To fulfill this requirement, the matrix containing the
roughness heights was obtained by a linear
transformation of the input matrix.
3. To simulate the non-Gaussian surface, the
skewness and kurtosis of the input sequence has
been established for the desired skewness and
kurtosis of an output sequence. Finally the non-
Gaussian random series has been generated by using
the Johnson translator system.
4. The developed algorithm can be used to
simulate manufacturing processes, abrasive wear or
running in phenomena. This kind of simulation can
be further incorporated into a particular stress
analysis for tribological designs or contact failure
predictions.


ACKNOWLEDGEMENT

This paper was realized with the support of
BRAIN Doctoral scholarships as an investment in
intelligence project, financed by the European
Social Found and Romanian Government.


REFERENCES

1. Bakolas V., 2003, Numerical Generation of
Arbitrarily Oriented Non-Gaussian Three-
Dimensional Rough Surfaces. Wear, 254, pp. 546-
554.
2. Bushan B., Kim T.W. and Cho Y.J., 2006, The
Contact Behavior of Elastic/Plastic non-Gaussian
rough surfaces. Tribology Letters, 22, pp. 1-12.
3. Creu S. Sp., 2006, Random Simulation of
Gaussian Rough Surfaces. Part 1- Theoretical
Formulations., Bul. IPI, LII (LVI), 1-2, pp. 1-17.
4. Creu S. Sp., 2006, The Influence of the
Correlation Length on Pressure Distribution and
Stresses State in Elastic-Plastic Rough Contacts,
IJTC-2006, paper 12339, San Antonio, TX, USA.
5. Elderton W.P. and Johnson N.L., 1969, System
of Frequency Curves. Cambridge University Press,
London.
6. Greenwood J.A., Wu J.J., 2001, Surface
Roughness and Contact: An Apology. Meccanica,
36, pp. 617-630.
7. Hill I.D., Hill R., Holder R. L., 1976, Fitting
Johnsons Curves by Moments. Applied Statistics,
25, pp. 180-189.
8. Hu Y.Z. and Tonder K., 1992, Simulation of 3-
D Random Rough Surface by 2-D Digital Filter and
Fourier Analysis., Int. J. Mach. Tools Manufact,
Vol. 32, pp. 83-90.
9. McCool J., 1986, Comparison Models for the
Contact of Rough Surfaces., WEAR, 107, pp. 37-60.
10. Patir N., 1978, A Numerical Procedure for
Random Generation of Rough Surfaces., Wear,
263-277.
11. Robbe-Valloire F., 2001, Statistical Analysis
of Asperities on a Rough Surface., Wear, 249, pp.
401-408.
12. Watson W. and Spedding T.A., 1982, The
Time Series Modelling of Non-Gaussian
Engineering Processes., Wear, 83, pp. 215-231.

ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 7-11
Cristina CIORNEI
email: tina_criss2005@yahoo.com
Emanuel DIACONESCU
email: emdi@fim.usv.ro
Department of Mechanical Engineering,
Stefan cel Mare University of Suceava,
ROMANIA
PRELIMINARY THEORETICAL SOLUTION FOR
ELECTRIC CONTACT RESISTANCE BETWEEN
ROUGH SURFACES
In contacts design, it is important to know the contact pressure, the
real contact area and the electrical contact resistance. This depends
on the material conductibility, on the geometry of the contacting
surfaces, on the applied load and on the current through the contact.
This paper aims to determine numerically, by CG-DCFFT
technique, contact area configuration and dimensions, in the case of
rough surfaces. Knowing the microcontact areas configuration and
dimensions, the electrical resistance is computed with analytical
formulas.
Keywords: numerical simulation, electrical contact resistance, CG-
DCFFT
1. INTRODUCTION
When electric current passes through a
contact, the size of the contact area has an important
influence on the contact resistance characteristics
due the constriction of the current lines at very small
contact areas. In theory, it was proved that up to the
nano-scale, the contact conductance, which is the
reverse of the electrical resistance, is proportional to
the contact domain perimeter. At nano-scale, the
contact conductance is proportional to the contact
area [1].
Analytical approaches of contact problems
are limited to a small number of contact geometries
and therefore numerical solution was imposed.
Since this requires meshes with large numbers of
nodes in the estimated contact domain,
unconventional fast numerical methods, such as the
multi-level multi-summation (MLMS) and the fast
Fourier transform (FFT) techniques, have been
developed. The best known algorithm was proposed
by Polonsky and Keer [2]. The most efficient
method in terms of computational effort combines
the Discrete Convolution Fast Fourier Transform
(DCFFT) algorithm with the conjugate gradient
(CG) method [3]. Creu [4] developed a fast and
original algorithm based on CG-FFT to study the
finite length line contact. Spinu [5] implemented a
CG algorithm similar to that proposed in [2], where
the MLMS routine was replaced with one based on
the DCFFT technique.
This study aims to find the shape and
dimensions of total contact area, as well as
individual micro-areas, using the CG-DCFFT
method combined with contact resistance calculus
for rough surfaces under normal loading.
2. FORMULATION
A contact between a curved rough surface
and a flat is considered. Coordinate system origin is
established in the common plane of contact, namely
the plane tangential to both bodies if they were
smooth enough and would initially form a point
contact. In this plane, analysis domain is divided
into elements of the same size, centered on grid
nodes.
To describe the initial contact, the geometry
was acquired using a 3D scanner. The obtained
data, namely heights associated to nodes of a
uniformly spaced rectangular grid with M lines and
N columns, represent the surface topography of the
equivalent punch. Consequently, punch geometry is
inserted as a matrix describing the digitized
topography of the rough surface.
The digitization of the equations and
inequalities which describe the elastic contact
problem lead to the following formulation [5]:
ij ij ij
r w z , (i, j) D; = + o e (1)
M N
ij k i k , j
k 1 1
w K p , (i, j) D;

= =
= e

(2)
M N
ij
i 1 j 1
Q ab p ;
= =
=

(3)
ij ij
r 0, p 0, (i, j) A; = > e (4)
ij ij
r 0, p 0, (i, j) D\ A; > = e (5)
where r is the gap between the deformed surfaces, w
is the total displacement in z-axis direction, z is the
initial contact geometry, K is the influence
8
coefficients matrix, p is the contact pressure, M N
is the number of grid nodes, a b is the area of the
elementary cell, A is the real contact area, D is the
analysis domain and Q is the static force. The
system is to be solved in pressures in the contact
area, namely the set of nodes in contact.
In the DC-FFT algorithm, which is efficient
in both computational time and storage, the linear
convolution is computed as a cyclic convolution.
The influence coefficients K(i k, j ) , which
represents the deflection of a node (i,j) due to a
uniform pressure acting on the rectangular element
(k, ), is obtained using closed-form expressions [6].
The influence coefficients matrix K, of size M N ,
is symmetric and positive definite, which leads to
application of methods like Steepest Descent or
Conjugate Gradient. If the mesh is uniform, K has
at most M N distinct elements. The extension of
the two members of convolution is made differently.
The pressure domain is extended with a ratio of two
in every direction, by maintaining the original
pressures in place and by completing the rest of
positions with zeros. This technique, called zero-
padding, differs from the one used for the influence
coefficients matrix, namely zero-padding and wrap-
around order, which is described in [5]. Then, p and
K are transferred from the space domain into
frequency domain, by applying a two-dimensional
fast Fourier transform to the extended matrix. The
domain extensions are removed and only the real
part of convolution is retained.
By combining the DCFFT technique with
conjugate gradient method, an efficient algorithm for
the resolution of pressure distribution and contact
area is obtained. Since the computational process is
iterative, a initial guess value for pressures is
required. The starting nodal pressures must be all
positive and must obey the static equilibrium
condition (3). In the case of electrical contacts,
where the load is applied centrically, the initial guess
value is the mean pressure acting on the potential
contact domain:
ij m
1 2
Q Q
p p , (i, j) D
MaNb L L
= = = e (6)
A distinctive feature of this scheme is that the
normal displacement is not computed during the
iterative process. In most contact solvers, the
displacement is subject to the outermost level of
iteration, in order to satisfy the force balance
equation. Here, this is imposed by updating the
pressure distribution at each iteration, according to
the relation between numerical and imposed load.
3. ELECTRICAL CONTACT RESISTANCE
The contact resistance is the electrical
resistance the current has to overcome when passing
through a closed contact. In the case of clean
metallic surfaces, electrical contact resistance is
defined of the constriction of the lines current, when
is forced to pass through a small contact area. For a
circular monocontact, the contact resistance is given
by Holms formula [7]:
c
R
2a

= , (7)
where is the contact material resistivity and a is
the contact radius.
For an elliptic monocontact, the expression
for resistance is:
c
b a
R U(m), m
(a b) b a

= =
t + +
, (8)
where U(m) is a elliptic integral of the first kind.
For a contact having a square area of side L,
the resistance is [8]:
c
R 0, 868
L

= , (9)
while for a rectangular contact, of width w and
length :
c
R 0, 868
w

=

. (10)
One can observe that contact resistance is
inversely proportional to the contact perimeter, not
to the contact area. These formulas are valid in case
of smooth surfaces. In real cases, the surfaces of the
contact elements are not smooth, but have an
inherent roughness. Therefore, a single contact is no
longer established between the two bodies, but a
multiple contact, formed by many spots created
between asperities.
Usually, the micro-contacts are made in the
form of revolution bodies, so that when two
elements are brought into contact, they do not form a
single point contact, but an assembly of individual
contacts. Under load, instead of contact nominal
surface, many individual contact areas will form. In
this case, current flows through contact micro-areas,
namely at their peripheries. The electrical contact
resistance is proportional to the radius of the contact
between the asperities. To determine the contact
resistance of such a contact, roughness distribution
is assumed to be homogeneous. The asperity tips
are assumed spherical and form an elementary Hertz
contact, of radius a, while R is the contact radius
assumed smooth and computed with Hertz formula.
Since these contacts are electrically independent, the
resistance of the micro-contacts is given by their
parallel resistance. Experimental tests show that the
contact resistance is bigger than parallel resistance
due to interactions between micro-contacts and lines
9
current flow distribution. The micro-contacts are
not independent, due to field pattern division. In
such a situation, the electrical contact resistance is
given by:
c
R
2na 2R n

= + , (11)
where n is the number of micro-contacts.
To obtain a smaller contact resistance, one
must act on the contact macro and micro-geometries.
Therefore, to achieve a uniform current distribution
on the contact area, it is required that the tip radius
and maximum pressure are the same on all
asperities. Maximum Hertz pressure depends on the
local load which is proportional to the mean contact
pressure, namely to the contact pressure between the
equivalent smooth surfaces. In order to obtain a
more uniform distribution of current density over
contact area, the pressure between equivalent
smooth surfaces must be as uniform as possible.
Contact pressure optimization is realised by
rounding the edge concentrators in a contact
between a flat ended rigid punch and an elastic half-
space. The pressure distribution is computed by a
simple numerical method [9]. Uniform contact
pressure can not be obtained for a flat equivalent
punch, but by using a curved surface, crowned
towards the middle.
These formulas are valid up to the micro and
submicroscopic scale, namely up to approximately
10 nm.
At nano-scale, the contacts behave
differently. A nano-contact is a contact between two
macroscopic bodies of a size comparable to electron
average mean free path; ballistic phenomena occur.
Usually, the size area of nano-contacts is less than
40 nm. Thus, at nano-scale the contact resistance is
given by Sharvins formula [10]:
c
2
4
R
3 a

=
t
, (12)
where is the electron average mean free path and
a is the transversal section radius. This formula is
valid only in the case where the contact size is
smaller than electron average mean free path.
4. RESULTS
The real surfaces were measured with an
optical profilometer. Using conversion of the
measured data in ASCII form, rough contact
geometry was inputted to the described numerical
program. Pressure distribution and real contact area
between a curved rough surface and a smooth plane
were obtained. The values of applied force ranged
from 0.01 N to 0.7 N. Electroplated gold micro-
contacts were used, whose curvature radii are less
than 1 mm. Figure 1 illustrates typical contact
pressures for a 255x255 [m x m] domain, meshed
in a 256x512 grid.
a.
b.
Figure 1. Pressure distribution at
0.1 N (a) and 0.4 N (b)
Figure 2 illustrates the variation of real
contact area at different loads. One can observe that
as force increases, the number of asperities brought
into contact also increases.
The numerical program yields the pressure
distribution, the number of micro-contacts, and their
shape and dimensions also. At the considered loads,
because the punch surface asperities have an
ellipsoidal shape and the meshed domain is divided
into rectangular elements, the micro-contacts are
also rectangular and distributed within the apparent
contact area, which is also rectangular in shape.
Therefore, equation (10) is employed to assess the
contact resistance. The global micro-contacts
resistance is found by summing parallel and
interaction resistances. Figure 3 illustrates the
obtained dependence of conductance on the real area
and perimeter.
10
Figure 2. Dependence between real contact area and applied force
Figure 3. Contact resistance dependence on contact
area and perimeter
Figure 4. Contact resistance dependence on contact
area and perimeter for elliptical micro-contacts
11
If micro-contacts are considered elliptical, the
resistance is computed using equation (8).
Conductance dependence on contact area and on
perimeter are depicted in Figure 4. In this case,
conductance has higher values.
5. CONCLUSIONS
The work reported herein can be summarized
by the conclusions reviewed below.
Theoretical investigations of the electrical
contact resistance show that its inverse counterpart,
the conductance, is proportional to the contact
circumference in macro and micro-contacts. In
nano-contacts, the conductance is proportional to the
contact area.
From a mechanical point of view, improving
or optimizing electrical contacts means altering
micro-asperity surfaces according to a polynomial
law, so that the micro-contact areas increase rapidly
with the load, leading to a low contact resistance
even at low loading levels.
The present numerical model can be used to
compute the shape and dimensions of total contact
area, as well as individual micro-areas for rough
surfaces under normal loading.
Reported results show that contact resistance
decreases when the load increases, and that contact
conductance depends linearly on the contact area
circumference, in agreement with general theory.
REFERENCES
1. Glonvea, M., 2006, Investigations upon micro
and nanocontacts with MEMS applications (in
Romanian), Research Report, Grant CNCSIS.
2. Polonsky, I.A., Keer, L.M., 1999, A
Numerical Method for Solving Rough Contact
Problems Based on the Multi-Level Multi-
Summation and Conjugate Gradient Techniques,
Wear, 231, pp. 206-219.
3. Grdinaru, D., 2006, Modelri numerice n
teoria contactului elastic (in Romanian), PhD
Thesis, Suceava, Romania.
4. Cre u, S., Antaluca, E., Cre u, O., 2003, The
Study of Non-Hertzian Concentrated Contacts by a
CG-DCFFT Technique, Rotrib03 National
Tribology Conference, Galai.
5. Spinu, S., Gradinaru, D., Marchitan, M.,
2006, FFT Analysis of Elastic Non-Hertzian
Contacts Effect of Rounding Radius upon Pressure
Distribution and Stress State, VAREHD 13,
Suceava.
6. Johnson, K.L., 1985, Contact Mechanics,
Cambridge University Press.
7. Holm, R., 2000, Electrical Contacts Theory and
Application, 4th edn, Springer Verlag, Berlin,
Heidelbrg, New York.
8. Braunovic, M., Konchits, V.V, Myskin, N.K.,
2006, Electrical Contacts. Fundamental,
Applications and Tehnology, CRC Press, Boca
Raton, London, New York.
9. Glonvea, M., Diaconescu, E., 2006,
Improvement of Punch Profiles for Elastic Circular
Contacts, Transactions of the ASME, Journal of
Tribology, Vol. 128, July 2006, 486 492.
10. Sharvin, Y.V., 1965, A Possible Method For
Studying Fermi Surfaces, Soviet Pysics Jetr, Vol.
21, pp. 65.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 12-18
Constantin-Ioan BARBIN
e-mail: costelbarbinta@yahoo.com
Spiridon CREU
e-mail: spcretu@mail.tuiasi.ro
Machine Design Department
Gheorghe Asachi Technical University Iasi
ROMANIA
THE INFLUENCE OF THE RAIL INCLINATION
AND LATERAL SHIFT ON PRESSURE
DISTRIBUTION IN WHEEL - RAIL CONTACT
Even though the UIC60 wheel profile and the S1002 rail are the
most used combination in the European rail transportation, the
interoperability is affected by the different rail inclination that
varies between the values of 1/40 and 1/20. The hunting motion and
the specific train motion in curve determine a permanently lateral
shift of the axle and, consequently, a permanent change of the initial
wheel-rail contact point. To find out the influence of these
modifications on pressure distributions, a fast and robust algorithm
has been used to solve the stress state in the general case of non-
Hertzian contacts. Brents method has been involved to find the
contact point for the unload conditions. To limit the pressure, an
elastic-perfectly plastic material has been incorporated into the
computer code.
Keywords: rail, wheel, lateral shift, rail inclination, pressure
distributions
1. INTRODUCTION
The running, as well as the reliability of the
wheel-rail unit, are based on the phenomena
developed within the concentrated contact loading.
Even though the UIC60 wheel profile and the
S1002 rail are the most used combination in the
European rail transportation, the interoperability is
affected by the different rail inclination that varies
between the values of 1/40 (Germany and Austria),
1/30 (Sweden) and 1/20 (France and Romania).
On the other hand, the hunting motion and
the specific train motion in curve determine a
permanently lateral shift of the axle and,
consequently, a permanent change of the initial
wheel-rail contact point.
The problem was first solved by Carter by
regarding the wheel-rail contact as a cylinder rolling
over a plane (a two-dimensional problem), (see
Ayasse, [1], Enblom, [2]).
Figure 1. The contact ellipse and ellipsoidal
pressure distribution (Hertzian) [2]
Three decades later, de Pater and Johnson,
(see Enblom, [2]), predicted the shape and size of
the contact area and pressure distribution
considering the Hertzian three-dimensional solution,
Figure 1.
In fact, the wheel-rail concentrated contact
appears as a non-Hertzian contact because of the
following violations of the Hertzian assumptions:
- the surfaces separation around the initial
contact point can not be expressed as a
quadratic form;
- the common generatrix has a finite length;
- the contacting surfaces are no longer smooth;
- friction is present on the contacted area.
Figure 2 points out the longitudinal and
lateral creepages accompanying the main rolling
loading.
Apart from the approximated solutions, the
general case for modeling the wheel-rail contact
must be solved numerically. Kalker was the first to
solve the general wheel - rail contact, for which he
developed the numerical program CONTACT,
Wiest [3].
For vehicle dynamics problems, where the
external contact parameters change continuously, i.e.
lateral position between wheel and rail profiles, the
program CONTACT cannot be used due to the high
computational time.
To overcome this, Kalker proposed a new
contact model called FASTSIM. A survey of these
methods is made in [2,3].
13
Finite element methods are also applied to the
wheel-rail contact problem and significant
simulations and developments have been recently
reported in literature, Damme [4].
Figure 2. Wheel-rail loads, [2]
The state-of-the-art papers of Knothe et al.
[5] discuss in more detail the above methods of
contact mechanics applicable for wheel - rail
contact.
More recent work on the elastic non-Hertzian
contact was made by Cretu [6,7], who solved the
contact between two randomly shaped bodies
described as half-spaces by using the Papkovici -
Boussinesq solution.
The developed numerical program is called
NON-HERTZ and its solving algorithm uses the
Conjugate Gradient Method involving the Discrete
Convolution with the process of zero padding and
wrap-around order associated with FFT.
Displacement is regarded as a convolution of
pressure and elastic response.
In the wheel-rail contact, the separation
between the contacting surfaces depends on a lot of
variables, as wheel and rail profiles, rail inclination,
track gauge, inside gauge and lateral shift of the
axle.
Figure 3. The real and hypothetical contact areas
2. NUMERICAL FORMULATIONS
A hypothetical rectangular contact area
denoted by
h
A is considered in the common tangent
plane, around the initial contact point. The
hypothetical area is large enough to overestimate the
unknown real contact area,
h r
A A > , Figure 3.
A Cartesian coordinate system (x, y, z) is
introduced, its xOy plane being the common tangent
plane, and with its origin located at the left corner of
the hypothetical rectangular area. The elastic
deflection of each surface is measured in the
direction of the corresponding outer normal and is
denoted by w
I
(x, y) and w
II
(x, y), respectively. The
sum of the individual deflections at any generic
point (x, y) is defined as a composite deflection,
denoted by w(x, y).
A uniformly spaced rectangular array is built
on the hypothetical rectangular contact area with the
grid sides parallel to the x and y-axes, Figure 3. The
nodes of the grid are denoted by (i, j), where indices
i and j refer to the grid columns and rows,
respectively. In the considered Cartesian system, the
coordinates of the grid node (i, j) are denoted by
(x
i
, y
j
) and are given by:
i
x i x = A , (0 i Nx) s < , (1)
and
j
y j y = A , ( 0 j Ny s < ), (2)
where x A and y A are the grid spaces in the x and
y-directions, respectively. The real pressure
distribution is approximated by a virtual pressure
distribution, a piecewice-constant approximation
between grid nodes being typically used, Figure. 4.
Figure 4. The real pressure distribution and
piecewice-constant approximation
14
The numerical formulation is given by the
following set of discrete equations:
a) the geometric equation of the elastic
contact:
ij ij ij ij 0
g h R w = + o ; (3)
b) the integral equation of the normal surface
displacement, (Boussinesq formula):
Ny 1 Nx 1
ij i k, j k
k 0 0
w K p


= =
=

, (4)
where the influence function K
ij
describes the
deformation of the meshed surface due to a unit
pressure acting in element (k, ), Cretu [7];
c) the load balance equation:
Ny 1 Nx 1
ij
i 0 j 0
x y p F

= =
A A =

, (5)
where F is the applied normal force.
d) the constraint equations of non-adhesion
and non-penetration:
ij
g 0, =
ij
p 0, >
r
(i, j) A e ; (6)
ij
g 0, >
ij
p 0, =
r
(i, j) A e ; (7)
e) the elastic-perfectly plastic behavior of the
material:
ij Y ij Y
p p p p > = , (8)
where
Y
p is the value of the pressure able to initiate
the plastic yielding.
The components of the stress tensor induced
in the point M(x,y,z) are obtained by superposition:
Ny 1 Nx 1
ij ijk k
k 0 0
(x, y, z) C p

= =
o =

, (9)
where the influence function
ijk
C (x, y, z)

describes
the stress component
ij
(x, y, z) o due to a unit
pressure acting in patch (k, ).
That is a Neumann type problem of the
elastic half-space theory. Closed form expressions
can be found in Hill [8].
3. ELASTIC-PERFECTLY PLASTIC SOLVER
A numerical algorithm has been developed to
solve the problems connected with the non-Hertzian
concentrated contact, Cretu [6]. The Conjugate
Gradient Method (CGM), with the iterative scheme
proposed by Polonsky and Keer, [9,10], has been
chosen to solve the mentioned algebraic system of
equations.
In order to increase the efficiency of the
numerical algorithm, a dedicated real discrete fast
Fourier transform routine for 3D contact problems
has been developed and incorporated into the code,
Creu [6], Nlias [11]. In the following the name
non-Hertz is used for the computer code.
This algorithm has been further applied to the
wheel-rail concentrated contacts and a solver code in
C++ language has been finally obtained. This solver
appears as a robust and fast alternative solution to
the finite element models that require large memory
and important computational resources, as well as to
the experimental tests which require expensive
equipments and very long duration.
By entering the input data (wheel profiles,
external normal load, wheel radius, yaw angle,
inside gauge, lateral shift of the axle, rail profiles,
rail inclination, track gauge, traction coefficient,
elasticity modulus, Poisson ratio etc) the pressure
distribution and the appropriate stress state for
various running conditions are obtained.
4. THE CONTACT GEOMETRY AND RIGID
SEPARATION
4.1 Rail and wheel
The wheelset and track gauges are shown
schematically in Figure 5. The track gauge is
measured between the points on the rail profile
located inside the track at a distance of 14.1 mm
from the common tangent to the profiles of both
rails. Assuming the track is in a straight line, the
mentioned tangent will be horizontal. The wheel
radius is measured at the mean wheel circle, usually
at 70 mm from the back of the wheel.
The two considered counterparts are a S1002
wheel profile and a UIC60 rail. The wheel has a
radius in rolling direction of 460 mm and the rail is
inclined at 1/40. The inner gauge of the wheelset is
1360 mm and the track gauge is standard, i.e. 1435
mm.
When the wheelset is in perfect alignment
with the track, the above dimensions would result in
a lateral shift between the left wheel and rail of 3
mm. Of course, during train movement, the wheelset
changes its relative position to the rail.
The standard UIC60 rail profile is defined by
arcs of circles and it is geometrically given as a
technical drawing. For keeping the same format as
for the wheel, the circles are approximated by
equations. Since the (not inclined) rail profile is
symmetrical, only half of it will be described in four
sections.
The standard S1002 divides the wheel profile
in eight sections; in each of these sections the profile
is defined by a specific algebraic polynomial.
15
Figure 5. Wheel-rail contact geometry
Table 1. Rail inclination
y 3 2 1 0 -1 -2 -3 -4 -5
yr 6.491 5.491 4.491 3.491 2.491 1.491 0.491 -0.509 -0.509
y
CR
11.96 12.485 13.07 13.745 14.54 15.545 16.925 26.27 26.72
0
y
CW
8.96 10.485 12.07 13.745 15.54 17.545 19.925 30.27 31.72
yr 6.063 5.063 4.063 3.063 2.063 1.063 0.063 -0.937 -1.937
y
CR
-4.21 -3.55 -2.44 -0.28 11.33 12.05 12.92 14.045 15.755
1
/
4
0
y
CW
-7.21 -5.55 -3.44 -0.28 12.33 14.05 15.92 18.05 20.76
yr 5.911 4.911 3.911 2.911 1.911 0.911 -0.089 -1.089 -2.089
y
CR
-7.345 -7.195 -6.877 -6.31 -5.395 -3.73 11.78 12.755 14.09
1
/
3
0
y
CW
-10.345 -9.195 -7.877 -6.31 -4.395 -1.73 14.78 16.755 19.09
yr 5.597 4.597 3.597 2.597 1.597 0.597 -0.403 -1.403 -2.403
y
CR
-10.915 -10.96 -10.99 -10.99 -10.96 -10.915 -10.825 -10.72 -10.585
R
a
i
l
i
n
c
l
i
n
a
t
i
o
n
1
/
2
0
y
CW
-13.915 -12.96 -11.99 -10.99 -9.96 -8.915 -7.825 -6.72 -5.585
An estimated target domain was meshed and
the separation matrix was used as input into the
NON-HERTZ code.
In the Table 1, y is the lateral shift of
wheelset relative to the track and yr is the lateral
position of the wheel relative to the rail. y
CR
is the
lateral coordinate of the contact point in rail
coordinates and y
CW
the lateral coordinate of the
contact point in wheel coordinates.
The standard notation and main dimensions,
involved in the contact geometry are as follows,
(Fig. 5):
- WM the middle of the mounted axle;
- TA the railway axis..
- track gauge: TG =1435 [mm];
- inside gauge: IG = 1360 [mm];
- wheel radius: Rw = 460 [mm];
- rail inclination: RI = 1/40;
- lateral shift of the axle: y = 0 [mm];
- yaw angle: 0;
- roughness amplitude: 0.0 [m];
- wheel profiles: S1002, are described by
polynomials;
- rail profiles: UIC60,are described by
polynomials.
4.2. The rigid contact separation
The separation h(x, y) between the
contacting surfaces depends on a lot of variables as
wheel profiles, rail profiles, rail inclination, track
gauge, inside gauge and lateral shift of the axle,
Figure 5.
The transversal positioning of the wheel
against the rail is achieved according to the
following equation:
yr y
U TG / 2 70 IG / 2 A = + A . (10)
The Brents method has been incorporated
into the computing scheme to find, for the unloaded
conditions, the first contact point of the two
surfaces. The Brents method combines root
bracketing, bisection, and inverse quadratic
interpolation to converge from the neighbourhood of
a zero crossing. The final form for the separation
h(x, y) was found as follows:
16
2 2
h(x, y) zw(y) rw(y) rw(y) zw(x) zr(y) = +
(11)
where zw(y) is wheel profile at the coordinate y,
rw(y) is the wheel radius at coordinate y, zw(x) is
the wheel profile at coordinate x, and zr(y) is the
rail profile at coordinate y.
The 2D profiles and 3D rigid separation
h(x, y) are exemplified in Figure 6.
4.3 Material properties and load:
- Young modulus: 2.1 10
5
[MPa];
- Poisson ratio: = 0.28;
- yield limit:
Y
p 580 = [MPa], corresponding
to R7T steel;
- external normal load 90 [kN].
a. 2D profiles
b. 3D rigid separation
Figure 6. Wheel-rail contact geometry (a) and 3D
rigid separation (b)
5. ELASTIC ANALYSIS
5.1 Elastic pressure distributions
The constraint (8) has not been involved in
the elastic analysis. The accuracy of the results
depends on the size of the uniformly spaced
rectangular array built on the hypothetical
rectangular contact area, Figure 7. The 3D pressure
distributions are exemplified in Figure 7a for an
array with 16x16=256 mesh points, and in Figure 7b
for an array with 512x512=262,144 mesh points.
The elastic conditions, normal loads and a lateral
shift s = 0 has been considered. The corresponding
2D distributions are plotted in Figure 8.
a.
b.
Figure 7. 3D pressure distributions
(elastic model, lateral shift s=0)
Figure 8. 2D pressure distribution
(elastic model, lateral shift s=0)
5.2. Influence of the lateral shift
The lateral shift of the wheel has a strong
influence on both shape of the real contact area and
maximum value of pressure distribution, as depicted
in Table 1 and in Figures 9 to 12.
17
Maximum contact pressure [MPa]
1614 1638 1624 1590 1544 1468 1337 2712 3748
3 2 1 0 -1 -2 -3 -4 -5
Lateral shift [mm]
Figure 9. Wheel S1002-Rail UIC60 with 0 inclination
Maximum contact pressure [MPa]
1030 964 873 932 1253 1450 1495 1428 2371
3 2 1 0 -1 -2 -3 -4 -5
Lateral shift [mm]
Figure 10. Wheel S1002-Rail UIC60 with 1/40 inclination (Germany, Austria)
Maximum contact pressure [MPa]
1161 1115 1060 995 913 880 1258 1435 1352
3 2 1 0 -1 -2 -3 -4 -5
Lateral shift [mm]
Figure 11. Wheel S1002-Rail UIC60 with 1/30 inclination (Sweden)
y
[
m
m
]
y
[
m
m
]
y
[
m
m
]
18
Maximum contact pressure [MPa]
1864 1839 1806 1778 1741 1701 1640 1565 1487
3 2 1 0 -1 -2 -3 -4 -5
Lateral shift [mm]
Figure 12. Wheel S1002-Rail UIC60 with 1/20 inclination (Romania, France)
5.3. Influence of the rail inclination
As shown in Figures 9 to 12, the rail
inclination appears to be a major factor influencing
the shape of the real contact area and, consequently,
the entire 3D elastic pressure distributions.
It can be noticed that a greater rail inclination
provides a greater maximum pressure.
6. CONCLUSIONS
1. The interoperability of the European rail
transportation is affected by the different rail
inclination that varies between 1/40 and 1/20.
2. A numerical solver has been involved to obtain
the 3D pressure distribution in non-Hertzian
wheel-rail contacts. This solver appears as a
robust and fast alternative solution to the finite
element models that require large memory and
important computational resources, as well as to
the experimental tests which require expensive
equipments and very long duration.
3. The lateral shift of the wheel alters considerably
both shape of the real contact area and maximum
value of the pressure distribution.
4. The rail inclination appears to be a major factor
influencing the shape of the real contact area
and, consequently, the entire 3D elastic pressure
distributions. Numerical simulations pointed out
that greater rail inclinations provide greater
maximum pressures.
REFERENCES
1. Ayasse, J. B., and Chollet, H., 2006, Wheel-
rail contact, in Handbook of Railway Vehicle
Dynamics, S. Iwnicki (Ed.), Taylor & Francis, pp.
85120.
2. Enblom, R., and Berg, M., 2008, Impact of
non-elliptic contact modelling in wheel wear
simulation, Wear, 265, pp 15321541.
3. Wiest M., Kassa E., Nielsen J.C.O., and
Ossberger H., 2008, Assessment of methods for
calculating contact pressure in wheel-rail/switch
contact, Wear, 265, pp. 1439-1445.
4. Damme, S., 2006, Zur Finite-Element-
Modellierung des stationren Rollkontakts von Rad
und Schiene, PhD thesis, Berichte des Instituts fr
Mechanik und Flchentragwerke Heft.
5. Knothe K., Wille R., and Zastrau, B., 2001,
Advanced contact mechanics-road and rail,
Vehicle System Dynamics, 35, (4-5), pp. 361-407.
6. Creu, S., 2005, Pressure distribution in
concentrated rough contacts, Bull.I.P. Iai, LI (LV),
1-2, pp. 1-31.
7. Creu S., 2009, Elastic-Plastic Concentrated
Contact, Iai: Polytehnium.
8. Hill, D. A., Nowell D., and Sackfield, A., 1993,
Mechanics of Elastic Contacts, Oxford: Butterworth.
9. Polonsky, I. A., and Keer, L. M., 1999, A
numerical method for solving contact problems
based on the multilevel multisumation and conjugate
gradient techniques. Wear, 231, pp. 206-219.
10. Polonsky, I. A., and Keer, L. M., 2000, Fast
methods for solving rough contact problems: A
comparative study. Trans. ASME, Journal of
Tribology., 122, pp 36-41.
11. Nelias D., Antaluc E., Boucly V., and Creu
S., 2007, A 3D semi-analytical model for elastic-
plastic sliding contacts, Trans. ASME, Journal of
Tribology, 129, pp. 671-771.
y
[
m
m
]
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 19-26
Cornel SUCIU
e-mail: suciu@fim.usv.ro
Emanuel DIACONESCU
e-mail: emdi@fim.usv.ro
Department of Mechanical Engineering,
University of Suceava,
ROMANIA
PRELIMINARY THEORETICAL RESULTS UPON
CONTACT PRESSURE ASSESSMENT BY AID OF
REFLECTIVITY
Several different experimental methods for investigating contact
features can be found in literature. The idea to optically investigate
the surfaces of contacting bodies [1-8], led to the development of a
new technique to measure the pressure distributions in a real contact
[9-11].
One of the contacting surfaces is covered, prior to contact
establishment, by a special gel. The contact closing removes the
excess gel and, during a certain time interval, the contact pressure
transforms the entrapped substance in an amorphous solid. In each
point, the refractive index of this solid depends on the pressure
acting during transformation. After contact opening, the reflectivity
of this coating depends on the former contact pressure and it is
mapped by aid of a laser profilometer, thus becoming an indicator
of contact pressure.
This paper studies the effect of pressure on the refractive index of
the solidified gel layer, as well as the different parameters that
influence its reflectivity. Using molecular physics and optics, a
theoretical model of reflectivity is studied and it is found to be
strongly influenced by both pressure and gel layer thickness. From
this model, pressure distribution laws are found for different ranges
of reflectivity and gel layer thickness.
Keywords: contact pressure, refractive index, reflectivity, solidified
gel layer
1. INTRODUCTION
Many different experimental methods can be
found in literature for the study of contact features.
The most advanced methods supply point to point
information on contact features, such as the
deformed surface of one or both contacting bodies,
or measurement of contact pressure and contact
stresses. An accurate method to find the deformed
surface of a metallic equivalent punch pressed
against a thick sapphire window as well as the actual
contact area was recently advanced by Diaconescu
and Glovnea [2-7] by aid of laser profilometry. By
using these experimental results as input data for
normal displacement, numerical calculations yield
the contact pressure responsible for these
deformations.
Yamaguchi, Uchida and Abraha [7] advanced
an interesting method of contact pressure evaluation,
based on measurement of intensity of a laser beam
reflected by the same surface, prior and after the
contact. They found that after contact the intensity
of reflected light increases in the former points of
contact area and become proportional to contact
pressure. Etching of the surface was found to
improve the methods sensibility.
Yamaguchi, Uchida and Abraha [8],
proposed a method for the assessment of contact
pressure distribution by means of a transferred oil
film. In this method, a thin film of oil is spread onto
the specimen and pressure is applied between this
surface and a clean, flat reference surface. Upon
releasing the load, part of the oil film is transferred
onto the measuring surface. The surface covered by
the transferred oil film is considered to be the real
contact area. The ratio of the area of the transferred
oil film to the apparent surface area is then detected
by the reflection of light.
The idea of Yamaguchi, Uchida and Abraha
to investigate optically the surface after contact was
opened led to the development of a new technique
for the evaluation of contact pressure in real contacts
[9-11]. This consists in measuring the reflectivity of
a thin coating formed on one of the contacting
surfaces as a result of transformation of a gel into an
amorphous solid at contact pressure.
As the refractive index of the coating depends
on the pressure inducing the change of phase, the
20
measured reflectivity is a useful indication of contact
pressure.
2. REFRACTIVE INDEX OF A SOLIDIFIED
GEL LAYER
As shown in the introduction, the
experimental method presented herein consists in
covering one of contacting surfaces with a special
molecular gel, prior to contact establishment. After
a well defined time, the contact is closed, the normal
load is applied and the system is maintained in this
state another adequate time interval. Although most
of the gel is expulsed at contact establishment, a
minute quantity remains at the interface of the two
contacting bodies. Under the action of the contact
pressure, the entrapped gel suffers a phase
transformation. Because the rate of pressure
increase to the nominal value is quite high, the
available time for molecular rearrangement in a low
viscosity state is short, of a few seconds only. The
gel viscosity, already high at contact establishment,
increases rapidly at contact pressure and impedes
molecular rearrangements. Consequently, the solid
state resulting from this transformation is an
amorphous one, and, therefore, isotropic. Finally,
the contact is opened and a very thin coating of
solidified gel is found on the previously covered
surface. This is an optical medium, characterized by
a refractive index.
The absolute refractive index is defined as the
ratio of the speed of the electromagnetic wave in
vacuum to the speed of the same wave when passing
the studied medium:
c
n
v
= = c , (1)
where: n real part of the refractive index; c speed
of light in vacuum; v speed of light in the studied
optical media; c dielectric constant or permitivity;
magnetic permeability.
Optical media can be transparent or
absorbing. A transparent medium has zero
conductivity and its magnetic relative permeability
differs from a unit value by a negligible amount.
Consequently, for such media, the refractive index
is:
n = c . (2)
No medium, except for vacuum, is perfectly
transparent. All material media show strong
absorption, at least in some regions of the
electromagnetic spectrum. An absorbing medium
has a finite conductivity and, consequently, a finite
current density. Nevertheless, in such media, the
volume charge density vanishes. The permittivity is
constant, but complex, because a phase-shift occurs
between the component field vectors. Similarly, the
conductivity is also complex. As a result, the
refractive index is complex, therefore symbolized by
n , the dielectric constant by c and the conductivity
by o . According to Ditchburn [12], the complex
refractive index is given by:
, ) n n 1 i n n i = _ = _ . (3)
Optically, the gel coating belongs to the class
of absorbing media and therefore its refractive index
is complex. The real part of this index depends on
pressure. If the value of the refractive index for a
reference pressure is known, its value at different
pressures is found by using the following equation
[9]:
2 2
r r r r
2 2
r r r r
2 2 n n
n
2 n n
+ 2 +
=
+ +
. (4)
If a dimensionless density, , defined as the
ratio of density at pressure p to that at reference
pressure is introduced in equation (4), the following
expression for the real part of the refractive index at
a given pressure p is obtained:
, )
, )
2 2
r r
2 2
r r
2 n 2 n 1
n
2 n n 1
+ +
=
+
. (5)
As shown in [13], the dimensionless density
can be expressed as a function of pressure if the
molecular interaction of specified substance is
known. To that end, it is first necessary to evaluate
energy for the crystalline lattice in the case of a
simple molecular crystal. To simplify the calculus,
this energy was determined for a perfect crystalline
lattice. It was assumed that molecular interactions in
such a lattice are governed by a Lennard-Jones-
London intermolecular potential, having the
following expression:
, )
12 6
r 4
r r
(
o o | | | |
= (
| |
\ \
(

, (6)
where r denotes the distance between the
interaction centers of observed molecules, o is the
value of r at which , ) r vanishes and is the
minimum value of the intermolecular potential.
Molecular lattice energy represents the
necessary work to extract a given molecule from the
lattice and to send it towards infinity. In fact, this
energy is equal to the half-sum of all potentials
between the observed molecule and all other
molecules in the crystal.
As shown by equation (6) of the Lennard-
Jones-London molecular interaction potential, this
21
potential decreases exponentially as the distance
between molecules increases. Thus, when
determining lattice energy for a certain molecule,
only molecules from neighboring layers have
important contributions, the effect of farther
molecules being negligible.
If the refractive index at atmospheric pressure
r
n , is known, and dimensionless density is
calculated as shown in [11], the refractive index at a
given pressure p can be determined using equation
(5). The refractive index of the solidified gel layer is
determined during solidification and depends on
contact pressure in each point. After contact
opening, the solidified gel layer retains contact
pressure distribution through its refractive index. As
this index varies along contact surface, gel layer
reflectivity becomes a function of applied contact
pressure.
Figure 1 illustrates how the real part of the
refractive index varies with increasing pressure
applied during solidification.
When a Hertz like pressure distribution is
applied the real part of the refractive index will vary
as shown in Figure 2.
0.2 5.9999999984 10
8
1.1999999999 10
9
1.7999999999 10
9
2.4 10
9
3 10
9

1.42
1.4208
1.4216
1.4224
1.4232
1.424
Solidified Gel Layer Refractive Index
Pressure [Pa]
R
e
f
r
a
c
t
i
v
e

I
n
d
e
x
n p ( )
p
Figure 1. Variation of the real part of solidified gel layer refractive index with increasing pressure
Figure 2. Refractive index variation along contact area, for Hertz pressure distribution
22
Since the reflectivity of the solidified gel
layer depends on the refractive index and the
extinction coefficient [9], it can be used as an
indicator of the pressure that occurred during contact
establishment.
Experimentally, reflectivity and gel layer
thickness are recorded using a laser profilometer and
used to determine the pressure distribution during
solidification.
3. REFLECTIVITY OF SOLIDIFIED GEL
LAYER
As shown in the introduction, solidified gel
layer reflectivity and thickness are mapped by laser
profilometry. When the laser beam meets the air
solidified gel layer interface, part of its energy
returns via reflection, while the rest traverses the
absorbent optical layer. Part of the incident energy
is lost by absorption, while the rest suffers a
reflection-refraction phenomenon at the gel layer
metal interface. Again, part of the light energy is
absorbed and part reflected. The reflected beam
traverses the gel layer, again being part reflected
part refracted at gel-air interface. When returning
into the air, the remainder of the beam energy
combines with the one first reflected by the gel
layer. The combined light wave is measured by the
laser profilometer, the ratio of incident light energy
to the reflected one yields the systems reflectivity.
This is a typical reflection refraction
problem, involving a three layer optical medium
having two optical interfaces, namely the air gel
layer interface and gel layer metal interface
respectively. As shown by Born and Wolf [14], at
each passing through one of these interfaces, the
incident laser beam is partly reflected and partly
transmitted, as shown in Figure 3. The process of
reflection refraction depends on the optical
properties of the two adjacent media.
Figure 3. Laser beam reflection-refraction when
passing through the solidified gel layer
Global reflectivity, measured by the laser
profilometer, is determined by several waves
reflected by the air-gel-metal optical system. The
gel surface reflectivity is given by a wave returning
from the surface
1
9 , given by the following
equation [9]:
, )
, )
2
2 2
2 2 2
1
2
2 2
2 2 2
n 1 n
n 1 n
+ _
9 =
+ + _
. (7)
This wave is then combined with a second
one,
2
9 , reflected by the gel metal interface after
passing through the gel layer. According to [9], this
second reflectivity can be calculated with:
, ) , )
, )
, )
, ) , )
2 2
2 3 2 3 3 2 2
2
2
2
2 2
2 2 2
2 2
3 2 3 3 2 2
16 n n n n n
1 n n
exp 4 d
.
n n n n
(
+ _ _

9 =
(
+ + _

o
+ + _ + _
(8)
The global, measured reflectivity is given by
the combination of the two waves, as follows:
2 2 2
1 2 1 2
9 = 9 +9 + 9 9 . (9)
Of great importance among the solidified gel
layer optical properties is its extinction coefficient.
If this coefficient is assumed constant in relation to
pressure, the phase shifting between the two waves
would remain constant, which is not the case.
Unfortunately, little information on the subject is
available in literature. Therefore, in order to find
theoretical profiles of reflectivity similar to those
measured experimentally, a relation between
extinction coefficient and pressure was adopted in
[9], based on experimental investigations:
, )
2
2 20
00
p
p 1 e
p
(
| |
( _ = _
|
(
\

, (10)
where
20
0.12 _ = is the extinction coefficient for the
gel layer solidified at atmospheric pressure,
00
p is
an important pressure, chosen equal to 5 GPa, and e
is a proportionality constant of 0.8 .
In the reflectivity equations presented above,
several notations were used, as follows:
2
n real
part of solidified gel layer refractive index, given by
either (4) or (5);
2
_ solidified gel layer extinction
coefficient, given by (10);
3
n metal refractive
index (considered to be
3
n 2.41 = in shown results);
3
_ metal extinction coefficient (considered to be
3
1.38 _ = in shown results); d gel layer
thickness, measured by laser profilometry; o
d
23
absorption coefficient of solidified gel layer, given
by:
2
2
2 n t
o = _

, (11)
where 780 nm = is the wavelength of the laser
beam used to scan the surface.
Both theoretical model and experimental
measurements obtained in [10-11], show that the
solidified gel layer global reflectivity is influenced
in each point by both solidification pressure and gel
layer thickness. As layer thickness was
experimentally found not to be constant along
contact area, its variation must be considered when
assessing contact pressure using reflectivity.
Figure 4.a depicts the theoretical variation of
global reflectivity with increasing pressure, for
various gel layer thicknesses between 0.1 m and
10 m . In Figure 4.b, the curves showing the
dependence of reflectivity on gel layer thickness
were traced at several constant pressures of
solidification.
5 10
8
9 10
8
1.3 10
9
1.7 10
9
2.1 10
9
2.5 10
9
2.9 10
9
3.3 10
9
3.7 10
9
4.1 10
9
4.5 10
9

15
19.5
24
28.5
33
37.5
42
46.5
51
55.5
60
Variatia reflectivitatii globale cu presiunea
Presiunea de solidificare
R
e
f
l
e
c
t
i
v
i
t
a
t
e
a

g
l
o
b
a
l

R p 0.110
6
,
, )
R p 0.510
6
,
, )
R p 1 10
6
,
, )
R p 5 10
6
,
, )
R p 10 10
6
,
, )
p
a)
0 1 10
6
2 10
6
3 10
6
4 10
6
5 10
6
6 10
6
7 10
6
8 10
6
9 10
6
1 10
5

15
19.5
24
28.5
33
37.5
42
46.5
51
55.5
60
Variatia reflectivitatii cu grosimea stratului de gel solidificat
Grosime strat
R
e
f
l
e
c
t
i
v
i
t
a
t
e
R 0.3 10
9
d ,
, )
R 0.5 10
9
d ,
, )
R 1 10
9
d ,
, )
R 2 10
9
d ,
, )
R 2.5 10
9
d ,
, )
d
b)
Figure 4. a) Reflectivity versus pressure, for several
gel layer thicknesses; b) Reflectivity versus gel
layer thickness, for several pressures
The surface illustrated in Figure 5 represents
global reflectivity variation when both pressure and
gel layer thickness are considered.
Variatia reflectivitatii cu presiunea si grosimea
R ( )
Figure 5. Global reflectivity variation with both
pressure and gel layer thickness
4. PRESSURE DISTRIBUTION ASSESSMENT
In order to assess pressure variation using
reflectivity, equation (9) must be solved with
pressure as an unknown. It was found that equation
(9) accepts solutions only for certain pairs of ranges
for gel layer thickness and global reflectivity. By
numerically solving this equation, for ranges of gel
layer thicknesses reflectivity values and the
corresponding values in reflectivity, pressure
variation curves were traced as shown in Figure 6.
0 1 10
6
2 10
6
3 10
6
4 10
6
5 10
6

0
1.2 10
9

2.4 10
9

3.6 10
9

4.8 10
9

6 10
9

Variatia presiunii
Grosime strat
P
r
e
s
i
u
n
e
LIN1
k1
LIN2
k2
LIN3
k3
LIN4
k4
d1
k1
d2
k2
, d3
k3
, d4
k4
,
Figure 6. Pressure variation with gel layer thickness
for different reflectivity values
By varying both reflectivity and gel layer
thickness when solving equation (9), several
corresponding pressure variation laws were
obtained, as illustrated by the surfaces traced in
Figures 7.a, 7.b, 7.c and 7.d respectively:
24
a) b)
c) d)
Figure 7. Pressure variation for several reflectivity and gel layer thickness ranges:
a) a gel layer thickness range of 2.7 3.2 m and a reflectivity variation between 32% and 40%;
b) a gel layer thickness range of 1.44 1.65 m and a reflectivity variation between 40% and 48%;
c) a gel layer thickness range of 0.53 0.96 m and a reflectivity variation between 50% and 55%;
d) a gel layer thickness range of 0.23 0.55 m and a reflectivity variation between 55% and 60%.
Figure 8. Pressure variation with reflectivity and gel
layer thickness
Figure 8 reunites in the same graph the
surfaces shown in Figure 7 in order to better
distinguish the different pressure variation laws
obtained by numerically solving equation (9).
Although it was found that solving equation
(9) with pressure as an unknown is only possible for
certain pairs of ranges in reflectivity and gel layer
thickness values, these ranges were found to be
consistent with practical application of the method,
as experimental measurements were contained
within these ranges.
Figure 9.c illustrates a typical pressure
distribution obtained when experimental data for
reflectivity (Figure 9.a) and corresponding gel layer
thickness (Figure 9.b) are taken into account when
solving equation (9). The shown experimental data
corresponds to a contact between a spherical
metallic punch and a metallic plate, with molecular
gel at the interface, as presented in [10-11].
25
0 20 40 60 80 100
60
66
72
78
84
90
Reflectivity profile
R
e
f
l
e
c
t
i
v
i
t
y

[
%
]
R1
k
R11
k
k
(a)
0 20 40 60 80 100
0
3 10
6

6 10
6

9 10
6

1.2 10
5

1.5 10
5

Gel layer profile

T
h
i
c
k
n
e
s
s
d1
k
d11
k
k
(b)
0 20 40 60 80 100
5.5 10
9

5.58 10
9

5.66 10
9

5.74 10
9

5.82 10
9

5.9 10
9

Pressure distribution
P
r
e
s
s
u
r
e
P1
k
P11
k
k
(c)
Figure 9. Reflectivity profile (a), corresponding gel
layer thickness (b) and pressure distribution (c), for
the contact between a metallic spherical punch and a
metallic plate
In Figure 9, the experimental data and
resulting real pressure distributions are traced with
dotted lines, and the continuous line represents the
approximation of the respective profiles when
disregarding the roughness effect.
Neither reflectivity profile, nor gel layer
thickness profile arent smooth because asperity
interactions generate steep peaks and deep valleys
with respect to ideal surfaces. At high resolutions,
the method can supply the shapes of reflectivity and
gel layer peaks and therefore yield asperity pressure.
5. CONCLUSIONS
The work reported herein can be summarized
by the conclusions reviewed below.
- Contact pressure assessment using
reflectivity is an experimental method based on the
solidification, inside the contact region, of a
molecular gel film applied on one of contacting
surfaces. The refractive index of the solidified gel,
as well as its extinction coefficient, depends on the
pressure acting during transformation, i.e. on contact
pressure.
- After contact opening, the reflectivity of the
surface initially covered with gel is scanned by aid
of a laser profilometer. Measured reflectivity
depends on refractive index, extinction coefficient
and local thickness of gel coating.
- The effect of solidification pressure upon
different optical properties of a gel layer (refractive
index, extinction coefficient etc.) was studied and
variation curves were traced.
- For a given set of molecular and optical
parameters, theoretical variation curves of
reflectivity were traced and its dependence on
pressure and on local gel layer thickness was
assessed.
- It was found that pressure has different
variation laws for different ranges of reflectivity and
of gel layer thickness.
- Experimental measurements of reflectivity
and corresponding solidified gel layer local
thickness were introduced in the numerical program,
thus obtaining real contact pressure distributions.
- Further research is needed to improve
accuracy of the method in order to find asperity
pressure distributions.
REFERENCES
1. Diaconescu, E.N., and Glovnea, M.L.,
Evaluation of Contact Area by Reflectivity, Proc.,
3rd AIMETA International Tribology Conference,
Italy, on CD, 2002
2. Glovnea, M.L., and Diaconescu, E.N., A New
Method for Experimental Investigation of Elastic
Contacts, (in Romanian), Symp. on Tradition and
Continuity in Railway Research, Vol.II, Bucharest,
1994, pp. 77-82.
3. Diaconescu, E.N., A New Tool for
Experimental Investigation of Mechanical Contacts,
Part I: Principles of Investigation Method,
VAREHD 9, Suceava, 1998, pp. 255-260.
4. Diaconescu, E.N. and Glovnea, M.L., A New
Tool for Experimental Investigation of Mechanical
Contacts, Part II: Experimental Set-Up and
Preliminary Results, VAREHD 9, Suceava, 1998,
pp. 261-266.
5. Diaconescu, E.N., and Glovnea, M.L.,
Validation of Reflectivity as an Experimental Tool
26
in Contact Mechanics, VAREHD 10, Suceava,
2000, pp. 471 476
6. Diaconescu, E.N., and Glovnea, M.L.,
Visualization and Measurement of Contact Area by
Reflectivity, Trans. of the ASME, J. of Trib., Vol.
128, october 2006, 915 917
7. Yamaguchi, K., Uchida, M., and Abraha, P.,
Measurement of Pressure on Contact Surface by
Reflection of Light (Effect of Surface Etching),
Proceedings of the Japan International Tribology
Conference, Nagoya, 1990, pp. 1271-1276.
8. Yamaguchi, K., Uchida, M., and Abraha, P.,
Measurement of the Pressure Distribution on
Contact Surfaces by the Detection of a Transferred
Oil Film, Surface Science 377379 (1997), 1015
1018
9. Diaconescu, E. N., Glovnea, M. L., Petroel,
O., A New Experimental Technique to Measure
Contact Pressure, Proc. of 2003 STLE/ASME Joint
International Tribology Conference, Ponte Vedra
Beach, Florida USA, 2003.
10. Suciu, C., Diaconescu, E., Spinu, S.,
Experimental Set-Up And Preliminary Results
Upon A New Technique To Measure Contact
Pressure, Proceedings of VarEHD14, Suceava, 9-
11 October, 2008, ISSN 1844-8917, Acta
Tribologica, vol. 16, 2008 ISSN 1220 - 8434.
11. Suciu, C., Diaconescu E., Contact Pressure
Assessement by Reflectivity of a Solidified Gel
Layer, Proceedings of the 2nd European
Conference on Tribology, ECOTRIB 2009, Faculty
of Engineering, Pisa, Italy, June 7 - 10, 2009
12. Ditchburn, R. W., Light, Vol. II, Blackie &
Son, Second Edition, 1963.
13. Diaconescu, E. N., 2004, Solid-Like Properties
of Molecular Liquids Subjected to EHD Conditions -
Theoretical Investigations, Proceedings of 2004
ASME/STLE International Joint Tribology
Conference, Long Beach, California USA, October
24-27,
14. Born, M., and Wolf, E., 1980, Principles of
Optics, Sixth Edition, Pergamon Press.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 27-33
Sergiu SPINU
e-mail: sergiu.spinu@fim.usv.ro
Department of Mechanical Engineering,
University of Suceava,
ROMANIA
NUMERICAL SIMULATION OF
ELASTIC-PLASTIC CONTACT
A fast algorithm for elastic-plastic non-conforming contact
simulation is presented in this paper. The plastic strain increment is
determined using a universal integration algorithm for isotropic
elastoplasticity proposed by Fotiu and Nemat-Nasser. Elastic-
plastic normal contact problem is solved iteratively based on the
relation between pressure distribution and plastic strain, until the
latter converges. The contact between a rigid sphere and an elastic-
plastic half-space is modeled using the newly proposed computer
program. Numerical simulations predict that residual stresses
decrease the peak intensity of the stresses induced by contact
pressure, thus impeding further plastic flow. Computed pressure
distributions appear flattened compared to elastic case, due to
changes in both hardening state of the elastic-plastic softer material
and contact conformity.
Keywords: elastoplasticity, plastic strain increment, effective
accumulated plastic strain, elastic-plastic contact
1. INTRODUCTION
While the elastic response of a material
subjected to load application is reversible, plasticity
theory describes the irreversible behavior of the
material in reaction to loading beyond the limit of
elastic domain. The transition between elastic and
plastic deformation is marked by the yield strength
of the softer material.
The modern approach in modeling elastic-
plastic contact is based on the algorithm originally
proposed by Mayeur, [1], for the elastic-plastic
rough contact. However, his implementation was
limited to two-dimensional contact, as influence
coefficients were derived for this case only.
Problem generalization is due to Jacq, [2], and to
Jacq et al. [3], who advanced a complete semi-
analytical formulation for the three-dimensional
elastic-plastic contact.
The algorithm was later refined by Wang and
Keer, [4], who improved the convergence of residual
and elastic loops. The main idea of the newly
proposed Fast Convergence Method (FCM) is to use
the convergence values for the current loop as initial
guess values for the next loop. This approach
reduces the number of iterations if the loading
increments are small. Wang and Keer used two-
dimensional Discrete Convolution Fast Fourier
Transform (DCFFT), [5], to speed up the
computation of convolution products.
Jacq's influence coefficients for residual
stresses were based on the problem decomposition
advanced by Chiu [6,7]. An alternative approach
was proposed by Liu and Wang, [8], who also
suggested that three-dimensional DCFFT can be
used in a hybrid algorithm incorporating convolution
and correlation with respect to different directions.
Their Discrete Correlation Fast Fourier Transform
(DCRFFT) algorithm uses convolution theorem to
assess correlation, by substituting one term of the
convolution product by its complex conjugate.
Nlias, Boucly, and Brunet, [9], improved the
convergence of the residual loop, by assessing
plastic strain increment with the aid of an algorithm
for integration of elastoplasticity constitutive
equations proposed by Fotiu and Nemat-Nasser,
[10], as opposed to existing formulation, based on
Prandtl-Reuss equations, [2]. As stated in [9], this
results in a decrease of one order of magnitude in the
CPU time.
Influence of a tangential loading in elastic-
plastic contact with isotropic hardening was
investigated by Antaluca, [11]. Kinematic
hardening was added to the model by Chen, Wang,
Wang, Keer, and Cao, [12], who advanced an
algorithm for simulating the three-dimensional
repeated rolling or sliding contact of a rigid sphere
over an elastic-plastic half-space.
Cretu and Benchea, [13], and Benchea and
Cretu, [14], employed an improved incremental
algorithm for elastic-plastic non-conforming contact
modeling, based on the method originally proposed
by Cretu and Hatmanu, [15]. This alternative
formulation uses an assumption of compatibility
28
between elastic and plastic strains, and can be used
to achieve accurate results with a moderate
computational effort, as implies fewer iterative
levels.
A numerical program for elastic-plastic
contact modeling is overviewed in this paper, and
the sequence used to assess the plastic strain
increment is presented in detail. The solver is used
to simulate the elastic-plastic contact between a rigid
sphere and an elastic-plastic half-space having a
hardening behavior described by Swift's law.
Numerical predictions agree well with results
obtained with alternative numerical codes or using
finite element analysis.
2. ELASTIC-PLASTIC CONTACT
ALGORITHM OVERVIEW
Since the works of Mayeur, [1] and of Jacq,
[2], Bettis reciprocal theorem is used in elastic-
plastic contact modeling to assess surface normal
displacement and stress state in an elastic half-space
in the presence of plastic strains. Resulting
equations suggest elastic-plastic contact problem
split in an elastic and a residual part. The elastic
part comprises the static force equilibrium,
interference equation, and complementarity
conditions, while the residual part expresses the
plastic strain increment and plastic zone contribution
to surface normal displacement and to stress field in
the elastic-plastic body. However, the two
subproblems cannot be solved independently, as
residual displacement, computed in the residual
subproblem, enters interference equation in the
elastic part, while contact stress, assessed in the
elastic subproblem, is needed to find the plastic
strain increment in the residual part.
Analytical resolution of resulting equations is
available for neither the elastic, nor the residual part,
as integration domains, namely boundary region
with tractions and plastic strain volume respectively,
not known a priori, are arbitrarily shaped.
Therefore, numerical approach is preferred.
The principle of numerical approach consists
in considering continuous distributions as piece-wise
constant on the cells of a three-dimensional grid
imposed in a volume enveloping integration
domains. With this formulation, integration in the
continuous model of the elastic-plastic contact
model is replaced by multi-summation of elementary
cells individual contributions, known from the
influence coefficients or the Green functions. As
these multi-summation operations are in fact
convolution and/or correlation products, spectral
methods are applied to speed up the computation.
The numerical model of the elastic part is
obtained from that corresponding to a normal elastic
contact problem completed with the residual term,
namely the residual displacement, which is
superimposed into the interference equation.
Consequently, the elastic subproblem can be treated
as an elastic contact problem with a modified initial
contact geometry. The most efficient solver is based
on the conjugate gradient algorithm advanced by
Polonsky and Keer, [16], tweaked with the DCFFT
technique for convolution evaluation.
In the same manner, the residual part is
reformulated numerically, by imposing digitized
plastic strain distribution and finite load increments.
Plastic strain contribution to normal surface
displacement is expressed as a two-dimensional
convolution, computed by two-dimensional DCFFT,
[2]. The problem of residual stresses induced in the
half-space by an arbitrary distribution of inelastic
deformations is solved following a method
originally suggested by Chiu [6,7]. The hybrid three
dimensional spectral algorithms newly proposed by
Spinu, [17], result in a dramatic decrease in
computational effort.
The algorithm proposed for simulation of
elastic-plastic non-conforming contact with isotropic
behavior is based on three levels of iteration.
The innermost level, which assesses plastic
strain increment, corresponds to the residual part,
and has a fast convergence, as described in the
following section. The second level adjusts contact
pressure and residual displacement in an iterative
approach specific to elastic contact problems with
arbitrarily shaped contact geometry.
The outermost level is related to the fact that,
unlike elastic solids, in which the state of strain
depends on the achieved state of stress only,
deformation in a plastic body depends on the
complete history of loading. This level applies the
load incrementally, until the imposed value is
reached.
The algorithm for solving one loading step in
the elastic-plastic normal contact problem is
summarized in Figure 1.
Figure 1. Elastic -plastic algorithm
29
Firstly, the elastic problem with modified
contact geometry hi is solved, yielding contact area
and pressure distribution p . The latter can be used
to assess elastic displacement field
pr
u and stress
field
pr
. These terms represent the elastic part of
displacement and stress, namely that part that is
recovered once loading is removed. The stresses
induced by pressure are used in the residual
subproblem, to assess plastic strain increment. The
algorithm, based on a method originally proposed by
Fotiu and Nemat-Nasser, [10], is discussed in detail
in the following section. The computed plastic
strain increment is used to adjust the achieved
plastic zone
p
. Once the volume with plastic
strains is known, residual parts of displacement,
r
u ,
and of stresses,
r
, can be computed. As opposed
to their elastic counterparts, terms
r
u and
r

express a potential state, that remains after contact

unloading, if no plastic flow occurs during contact
relief. The total displacement can then be computed,
pr r
+ u u , thus imposing a new interference equation
in the elastic subproblem. These sequences are
looped until convergence is reached.
3. PLASTIC STRAIN INCREMENT
According to general theory of plasticity,
plastic flow occurrence can be described
mathematically with the aid of a yield function,
assessing the yield locus in the multidimensional
space of stress tensor components. If von Mises
criterion is used to express stress intensity, this
function can be expressed as:
p p
VM Y
f (e ) (e ) = o o , (1)
where
p
e denotes the effective accumulated plastic
strain,
p p p
ij ij
e 2 3 = c c , and
p
Y
(e ) o is the yield
strength function. The latter satisfy the relation for
the initial yield strength
Y0
o :
Y Y0
(0) o = o . (2)
The following conditions must be met:
p p
f 0; de 0; f de 0 s > = , (3)
with conditions f 0 = and
p
de 0 > corresponding to
plastic flow.
For elastic-perfectly plastic materials, relation
(2) is verified for any value of
p
e . However, for
metallic materials, more complex models of elastic-
plastic behavior are used, as the isotropic, or the
kinematic hardening laws. The isotropic hardening
law of Swift,
p p n
Y
(e ) B(C e ) o = + , (4)
with B, C and n material constants, is used in the
current formulation.
According to flow rule, plastic strain
increment can be expressed as:
ij p p p
ij
ij VM
3S
f
d de de
2
o
c = =
oo o
, (5)
where
ij
S denotes the deviatoric stress tensor.
The algorithm used to derive the plastic strain
increment was advanced by Fotiu and Nemat-
Nasser, [10], who developed a universal algorithm
for integration of elastoplasticity constitutive
equations. As stated in [10], the algorithm is
unconditionally stable and accurate for large load
increments, as it takes into account the entire non-
linear structure of elastoplasticity constitutive
equations, which are solved iteratively, via Newton-
Raphson numerical method, at the end of each
loading step. The yield function f is linearized at
the beginning of the load increment, by employing
an elastic predictor. This places the state point far
outside the yield surface f 0 = , since elastic-plastic
modulus is small compared to the elastic one. The
return path to the yield surface is generated by the
plastic corrector, via Newton-Raphson iteration.
This approach, also referred to as elastic predictor -
plastic corrector, is efficient when most of the total
strain is elastic. In the fully plastic regime, which
occurs usually after the elastic-plastic one, the
plastic strain is predominant, thus the return path
may require numerous iterations. Thus, linearization
at the beginning of the loading step is performed by
a plastic predictor, and return path is generated with
an elastic corrector.
A yield occurs when von Misses stress
exceeds current yield stress, namely when f 0 > .
The elastic domain expands or translates to include
the new state point, namely to verify condition
f 0 = . The actual increment of effective
accumulative plastic strain should satisfy equation of
the new yield surface in the plastic zone:
p p
f (e e ) 0 + o = . (6)
Here,
p
e o denotes the finite increment of
effective plastic strain, as defined in [2]. Relation
(6) can be considered as an equation in
p
e o , which
is solved numerically by Newton -Raphson iteration.
To this end, yield surface relation is linearized along
plastic corrector direction:
p
p p p p
p
f (e )
f (e e ) f (e ) e 0
e
c
+ o = + o =
c
, (7)
30
yielding the plastic corrector:
p p
p
p p
VM Y
p p p
f (e ) f (e )
e
f (e ) (e )
e e e
o = =
co c co

c c c
. (8)
For isotropic hardening, the derivate of
equivalent von Mises stress with respect to effective
accumulative plastic strain was derived by Nlias,
Boucly and Brunet, [9], from the general equations
presented in [10] for rate-dependent elastoplasticity:
VM
p
3G
e
co
=
c
. (9)
where G is the shear modulus, or the Lams
constant.
With these results, the following return-
mapping algorithm with elastic predictor - plastic
corrector can be formulated:
1. Acquire the state at the beginning of the
loading step and impose the elastic predictor. For
elastic-plastic contact problems, this is equivalent to
solving an elastic loop without imposing any
residual displacement increment. Corresponding
parameters are identified by an "a" superscript, as
opposed to a "b" superscript, used to denote the state
at the end of the load increment:
p(a)
e ,
(a) p(a)
Y Y
(e ) o = o ,
(a)
ij
o (
(a) pr(a) r(a)
ij ij ij
o = o + o ),
(a)
VM
o ,
(a) (a) (a)
VM Y
f = o o . These variables also represent the
input for the Newton-Raphson iteration. Thus, by
using superscripts to denote the Newton-Raphson
iteration number,
p(1) p(a)
e e = ,
(1) (a)
Y Y
o = o ,
(1) (a)
ij ij
o = o ,
(1) (a)
VM VM
o = o ,
(1) (a)
f f = .
2. Start the Newton-Raphson iteration.
Compute the plastic corrector according to relations
(8) and (9):
p(i)
p(i) (i)
p(i)
k(e )
e f 3G
e
| |
c
o = +
|
c
\ .
. (10)
3. Use the plastic corrector to adjust model
parameters:
(i 1) (i) p(i)
VM VM
3G e
+
o = o o ; (11)
p(i 1) p(i) p(i)
e e e
+
= + o ; (12)
(i 1) p(i 1)
Y Y
(e )
+ +
o = o ; (13)
(i 1)
(i 1) (1) VM
ij ij
(1)
VM
S S
+
+
o
=
o
. (14)
4. Verify if Eq. (6) is verified to the imposed
tolerance eps . If condition
(i 1) (i 1) (i 1)
VM Y
f eps
+ + +
= o o > (15)
is satisfied, go to step 2. If else, convergence is
reached, and the state point at the end of the loading
step is described by the newly computed parameters:
p(b) p(i 1)
e e
+
= ,
(b) (i 1)
VM VM
+
o = o ,
(b) (i 1)
ij ij
S S
+
= .
5. Compute the plastic strain increment,
according to Eq. (5):
, ,
(b)
ij p p(b) p(a)
ij
(b)
VM
3S
e e
2
oc =
o
. (16)
This increment is used to update the plastic
zone. The residual parts of displacement and of
stress can then be computed, and superimposed to
their elastic counterparts.
4. NUMERICAL SIMULATIONS
The contact between a rigid sphere of radius
6
R 105 10 m

= and an elastic-plastic half-space is

modeled, allowing for comparison with results
published by Boucly, Nlias and Green, [18].
Elastic half-space parameters are: Young modulus,
2
E 210GPa = , Poisson's ratio,
2
0, 3 v = . The
hardening law of the elastic-plastic material is
chosen as a power law (Swift), according to El
Ghazal, [19]:
p p n
Y
(e ) B(C e ) o = + , (17)
with
p
Y
(e ) o the yield strength function,
p
e the
effective accumulated plastic strain, expressed in
microdeformations, and the following parameters:
B 1280MPa = , C 30 = , n 0.085 = . The imposed
hardening curve is depicted in Figure 1.
Figure 2. Swift's hardening law
The contact is loaded incrementally to a
maximum value of W 0.650N = , for which the
purely elastic model predicts a contact radius
H
a 6.053 m = and a hertzian pressure
H
p 8470 MPa = .
Dimensionless coordinates are defined as
ratios to
H
a ,
i i H
x x a = , and dimensionless
31
pressure or stresses as ratios to
H
p . The
computational domain is a rectangular cuboid of
sides
1 2 H
L L 3a = = ,
3 H
L 1.6a = , which is
dicretized into
1 2
N N 120 = = ,
3
N 80 = elementary
cells. Due to the fact that problem is axisymmetric,
three dimensional distributions are depicted in the
plane
2
x 0 = only.
Pressure profiles predicted by the numerical
program for six loading levels corresponding to
elastic-plastic domain are presented in Figure 3.
Hertz pressure corresponding to maximum load is
also plotted for reference.
Figure 3. Pressure profiles in the plane
2
x 0 = ,
various loading levels
Plastic strain region generates residual
displacements, namely displacements that would
persist if a purely elastic unloading would occur.
These displacements increase contact conformity, as
depicted in Figure 4, resulting in a more uniform
pressure distribution on an enlarged contact area.
The curve describing the residual print maximum
depth versus loading level, Figure 5, is the same
with the one obtained by Jacq. et al, [3], and by
Benchea and Cretu, [20].
Figure 4. Residual print profiles in elastic-plastic
spherical contact
Figure 5. Residual print depth
Effective accumulated plastic strain
distribution is presented in Figure 6. Plastic zone,
initially occupying a hemispherical region located at
hertzian depths, advances toward the free surface
with increased loading, enveloping a purely elastic
core. This development is consistent with the one
suggested by Johnson, [21].
Figure 6. Effective accumulated plastic strain
Plastic strains induce residual stresses,
namely elastic stresses which persist in the elastic-
plastic body after contact unloading. These residual
stresses superimpose stresses induced by contact
pressure, the resulting state being responsible for
further plastic strain. Consequently, an accurate
estimation of stress field in the elastic-plastic body is
essential to plastic strain increment prediction.
Figures 7 and 8 depict distributions of
equivalent von Mises residual and contact stress.
Summation of these two states yields elastic-plastic
stress state, presented in Figure 9. To allow for a
better comparison, the same scale is used in Figures
7-9. Residual stress intensity is one order of
magnitude smaller than equivalent contact stress.
Comparison of distributions depicted in Figures 8
and 9 suggests that residual stress reduces peaks in
contact stress intensity, thus making the resulting
field more uniform. This behavior is also suggested
32
by the curves traced in Figure 10. Maximum
intensity of contact stress increase more rapidly than
the maximum of the total field, due to contribution
of residual stress. Consequently, residual stresses
act to impede further plastic flow, until the new
stress point reaches the new yield locus.
Figure 7. Von Mises residual stress
Figure 8. Von Mises contact stress
Figure 9. Resulting Von Mises stress in
elastic-plastic body
Figure 10. Maximum intensities of stress fields
versus loading level
5. CONCLUSIONS
This paper summarizes an efficient algorithm
for simulation of elastic-plastic contacts.
Computation of plastic strain increment is presented
in detail. The plastic strain increment is determined
in a fast convergent Newton-Raphson procedure
which iterates a scalar, namely the effective
accumulative plastic strain. The method, originally
suggested by Fotiu and Nemat-Nasser, employs an
elastic predictor, which places the state outside yield
surface, and a plastic corrector, assessing the return
path to the yield locus. The method is fast, stable
and accurate even for large loading steps.
Plastic strain modifies contact pressure by
superimposing induced residual surface
displacement into the initial contact geometry.
Contact pressure, in its turn, contributes to the
subsurface stress state, responsible for plastic zone
development. Consequently, the model is solved
iteratively based on the relation between pressure
distribution and plastic strain, until the latter
converges.
The newly proposed elastic-plastic contact
program is used to simulate the contact between a
rigid sphere and an elastic-plastic half-space
following a Swift isotropic hardening law. Pressure
profiles predicted numerically agree well with
already published results. Pressure appears flattened
compared to the elastic case, due to changes in both
hardening state of the softer material and contact
conformity.
Plastic zone, initially occupying a
hemispherical region located at hertzian depths,
advances toward the free surface with increased
loading, enveloping a purely elastic core. Residual
stress intensity is one order of magnitude smaller
than stresses induced by contact pressure. They
contribute to the total elastic field by decreasing the
peaks in contact stress intensity, thus impeding
33
further plastic flow. The resulting field is more
uniform, suggesting that material responds by
change in hardening state as to delay further plastic
flow until the state reaches the new yield locus.
ACKNOWLEDGEMENT
This paper was supported by the project
Progress and development through post-doctoral
research and innovation in engineering and applied
sciences PRiDE - Contract no. POSDRU/
89/1.5/S/57083, project co-funded from European
Social Fund through Sectorial Operational Program
Human Resources 2007-2013.
REFERENCES
1. Mayeur, C., 1995, Modlisation du contact
rugueux lastoplastique, Ph.D. Thesis, INSA Lyon,
France.
2. Jacq, C., 2001, Limite d'endurance et dure de
vie en fatigue de roulement du 32CrMoV13 nitrur
en prsence d'indentations, Ph.D. Thesis, INSA
Lyon, France.
3. Jacq, C., Nlias, D., Lormand, G., and
Girodin, D., 2002, Development of a Three-
Dimensional Semi-Analytical Elastic-Plastic Contact
Code, ASME J. Tribol., 124, pp. 653667.
4. Wang, F., and Keer, L. M., 2005, Numerical
Simulation for Three Dimensional Elastic-Plastic
Contact With Hardening Behavior, ASME J.
Tribol., 127, pp. 494502.
5. Liu, S. B., Wang, Q., and Liu, G., 2000, A
Versatile Method of Discrete Convolution and FFT
(DC-FFT) for Contact Analyses, Wear, 243 (12),
pp. 101111.
6. Chiu, Y. P., 1977, On the Stress Field Due to
Initial Strains in a Cuboid Surrounded by an Infinite
Elastic Space, ASME J. Appl. Mech., 44, pp. 587
590.
7. Chiu, Y. P., 1978, On the Stress Field and
Surface Deformation in a Half Space with Cuboidal
Zone in Which Initial Strains Are Uniform, ASME
J. Appl. Mech., 45, pp. 302306.
8. Liu, S. Wang, Q., 2005, Elastic Fields due to
Eigenstrains in a Half-Space, ASME J. Appl. Mech.,
72, pp. 871878.
9. Nlias, D., Boucly, V., and Brunet, M., 2006,
Elastic-Plastic Contact Between Rough Surfaces:
Proposal for a Wear or Running-in Model, ASME J.
Tribol., 128, pp. 236 - 244.
10. Fotiu, P. A., and Nemat-Nasser, S., 1996, A
Universal Integration Algorithm for Rate-Dependant
Elastoplasticity, Comput. Struct., 59, pp. 1173
1184.
11. Antaluca, E., 2005, Contribution a l'tude des
contacts lasto-plastiques - effet d'un chargement
normal et tangentiel, Ph.D. Thesis, INSA Lyon,
France.
12. Chen, W. W., Wang, Q. J., Wang, F., Keer,
L. M., and Cao, J., 2008, Three-Dimensional
Repeated Elasto-Plastic Point Contacts, Rolling, and
Sliding, ASME J. Tribol., 75, pp. 021021-1 -
021021-12.
13. Cretu, S. Sp., Benchea, M., 2008, An
Improved Incremental Model to Analyse Elastic-
Plastic Concentrated Contacts, Proc. of 16th
International Colloquium Tribology, Esslingen,
Germany, pp.33 (on CD also).
14. Benchea, M., Cretu, S. Sp., 2007, A Three
Dimensional Elastic Plastic Analysis of Rolling
Contacts, ROTRIB-07, Nov. 6-9, Bucharest,
Romania.
15. Cretu, S., Hatmanu, V., 1985, A Numerical
Analysis of Permanent Deformation in Elastic-
Plastic Line Contact, Bul. Inst. Polit. Iasi, XXXI,
(1-4), pp. 19-25.
16. Polonsky, I. A., and Keer, L. M., 1999, A
Numerical Method for Solving Rough Contact
Problems Based on the Multi-Level Multi-
Summation and Conjugate Gradient Techniques,
Wear, 231(2), pp. 206219.
17. Spinu, S., 2009, Contributions to the Solution
of the Elastic-Plastic Normal Contact Problem, (in
Romanian), PhD Thesis, University of Suceava,
Romania.
18. Boucly, V., Nlias, D., and Green, I., 2007,
Modeling of the Rolling and Sliding Contact
Between Two Asperities, ASME J. Tribol., 129, pp.
235 - 245.
19. El Ghazal, H., 1999, Etude des proprietes
microstructurales et mecaniques des aciers
16NiCrMo13 cemente et 32CrMoV13 nitrure -
Application a la prevision de leur limite
dendurance en fatigue de roulement, Ph.D. Thesis,
INSA Lyon, France.
20. Benchea, M, and Cretu, S., 2008, An
Improved Incremental Model to Analyse Elastic -
Plastic Concentrated Contacts The Finite Element
Analysis and Validation, Acta Tribologica, Vol. 16,
ISSN 1220-8434.
21. Johnson, K. L., 1985, Contact Mechanics,
Cambridge University Press.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 34-41





Yuichiro NAGATA
e-mail: y.nagata@sussex.ac.uk

Romeo GLOVNEA
e-mail: r.p.glovnea@sussex.ac.uk


School of Engineering and Design,
University of Sussex,
UNITED KINGDOM

DIELECTRIC PROPERTIES OF GREASE
LUBRICANTS

Grease lubricants are often preferred in machine elements working
under elastohydrodynamic conditions such as rolling element
bearings and constant velocity joints due mainly to the fact that they
do not need resupply and filtering systems and provide lower
lubricant losses. At the same time grease are complex materials,
with liquid and solid phases, which make difficult the prediction of
their behaviour in those contacts. In the present paper dielectric
properties of lubricating greases are studied, as a step towards the
understanding of their behaviour and predicting their in-contact
rheological properties.
Keywords: elastohydrodynamic lubrication, grease, dielectric
properties, rheology


1. INTRODUCTION

Elastohydrodynamic lubrication regime
(EHD) occurs in machine elements such as rolling
element bearings, gears, cams, traction drives and
others. The non-conformal contacts of these
machine elements are characterized by very thin
lubricating films (typically under one micrometer)
which support extremely large pressures due to a
combined effect of hydrodynamic action of the fluid,
the elastic deformation of the surfaces and the
variation of lubricant viscosity with pressure. In the
case of oil-lubricated EHD contacts there are
currently techniques for measuring experimentally
or predicting theoretically, with sufficient accuracy,
the film thickness. Optical interferometry on one
hand and various numerical techniques on the other
have been used in the past to successfully evaluate
the behaviour of elastohydrodynamic films in both
steady state and transient conditions [1]. The same
cannot be said about grease lubricated contacts,
where their dual, liquid-solid phase makes it difficult
to correlate their bulk properties to their film
forming capabilities.
The current paper is part of a wider study on
the rheology of lubricating greases in EHD contacts
and is focused on the correlation between their
dielectric characteristics and their in-contact
rheological properties.


2. BACKGROUND

Greases are very common lubricants in
rolling element bearings where they can provide a
life-long solution for these machine element
lubrication needs. However it is difficult to predict
the behaviour inside the contact or the film thickness
based on greases bulk properties as it is the case
with liquid lubricants.
Film thickness measurements by optical
interferometry have revealed that in general greases
give thicker film than their base oils at low speed,
but the dependence of the film thickness on the
entrainment speed is similar to that found for oils.
On the other hand, at greater speeds, the film
thickness decreases sharply due to starvation [2].
Errikson et al [3] have shown that soap particles are
able to pass through the EHD conjunction in a
certain proportion, not necessarily related to their
weight percentage in the grease composition. The
effect of the thickener on the film thickness has also
been studied by Couronne et al [4]. They have
found that thickener microstructure is not a
determinant factor for the formation of a thick
lubricant film, however, it influences oil bleeding,
mechanical stability, and rheological behaviour.
Cann et al suggested that the film thickness of grease
lubricated contacts increases approximately linearly
with soap concentration [5].
The rheological behaviour of lubricants in
elastohydrodynamic conditions has been a debated
subject for many years and a number of models have
been proposed by various researchers in the field.
The near exponential increase of the viscosity of the
lubricant with pressure has led to the conclusion that
this behaves like a visco-elastic solid with a
response depending on pressure, temperature and
shear rate.
Hirst and Moore [6] showed, based on
extensive experimental traction results, that the EHD
films behave elastically at low shear rates, but lose
elastic properties at higher rates of shear, when they
behave like viscous, non-Newtonian fluids. They
propose a traction coefficient (T/W) of the form:
0 0
0
0
T
ln
W p 2
| | t t
= ot

q
\ .

|
, (1)
where o is the pressure viscosity coefficient,
0
t is a
characteristic shear stress, is the viscosity at
ambient pressure and
0
q
is the shear rate.
Bair and Winer [7] suggest a modified
Maxwell model which allows for a limiting shear
stress of the lubricant film, based only on primary
laboratory results. Evans and Johnson [8]
investigated the behaviour of a number of
lubricating oils with different chemical structure
using a disc machine as a high-pressure rheometer.
By extending a previous model proposed by Johnson
and Tevaarwerk [9] they suggest that the behaviour
of lubricants in EHD conditions can be described by
the following relationship:
0
0
sinh
G
| | t t
= +
|
q t
\ .

t
. (2)
Diaconescu [10] has showed theoretically
that molecular liquids possess solid-like behaviour
when subjected to short duration shear, exhibiting a
shear modulus and limiting shear stress.
From an experimental point of view the film
formation in elastohydrodynamic contacts has been
investigated by either optical or electrical methods.
The former are mainly based on optical
interferometry requiring that one of the contacting
bodies is transparent, while in the latter, either the
resistance or capacitance of the contact between two
metallic bodies is analysed. Electrical methods are
relatively simple to implement and inexpensive but
are only able to give average values of the measured
parameters. When used to evaluate film thickness
they are also difficult to calibrate.
The electrical capacitance has traditionally
been used to measure film thickness in various EHD
lubricated system, such us piston-ring of IC engines
[11-13] and cam-tapet and gears mechanisms [14-
16]. Electrical capacitance depends of the frequency
of the electrical current passed through it because
the electric dipoles of the molecules need time to
align with the electric field. This property is
exploited in an experimental technique called
dielectric spectroscopy or dielectric relaxation
spectroscopy which is able to correlate molecular
processes with the rheological behaviour of a
sample.
In this paper dielectric spectroscopy has been
used in parallel with traction measurements to
analyse rheological parameters of various
composition grease lubricants.
3. EXPERIMENT

3.1 Experimental setup and procedure
Two types of tests were carried out, denoted
here as static and dynamic. Details and schematics
of the experimental setups of the static
measurements can be found in [17]. In these, a plane
capacitor has been set up between the jaws of a
micrometer, which allowed the separation between
plates to be set with a micrometer precision. The gap
between the plates was set at 0.2 millimetres in these
tests. After the capacitance of this capacitor in air
was measured, the space between the plates was
filled with grease and the capacitance measured
again. Capacitance has been measured in a
frequency range of 100Hz to 10MHz by an
impedance phase shift analyser. In this study an
extra resistor, 500k, was connected in series with
the capacitor so that the dielectric constant c and the
loss factor c were calculated from the measured
capacitance C, resistance R and the circular
frequency by using the following equations:
( ) ( )
2
2 2
0 0
C R
,
C 1 RC C 1 RC
e
' '' c = c =
C
(
+ e + e
(

. (3)
Although non-polar substances do not show
relaxation behaviour, the introduction of the external
resistor made possible that all samples show
relaxation behaviour. To calculate the relaxation
time
r
t of the samples, the Havriliak-Negami
equation (4) was fitted to the obtained data between
1kHz and 100kHz.
( )
(
0
r
i 0
1 i

|
o
c c
' ''
) , 1 c c = s o | s
(
+ et

, (4)
where
0
c and

c are the relative permittivity at low

frequency and at high frequency respectively.


Steel ball
Steel disc
Friction force
measuring system Loading
Motor


Figure 1. Traction coefficient measurement setup

35
Table 1. Composition of tested greases


EHD traction measurements, called here
dynamic tests, were performed in rig normally used
for measuring elastohydrodynamic film thickness. In
this study the EHD contact was formed between a
flat steel disc and a ball loaded together and driven
at appropriate velocities, such that desired slide/roll
ratios were obtained. The grease under study was
spread onto the surface of the disc and was not re-
supplied during the test. A schematic of the rig is
shown in Figure 1. The normal force was measured
by a load cell within the loading system, while the
traction force was measured by strain gauges placed
on a thin plate which drove the balls shaft.

3.2 Materials and testing conditions
Ten different types of grease were employed
in this study with their detailed composition listed in
Table 1. Greases 1 to 6 are made by NSK for
research purposes, containing no additives, solely to
see the effect of the base oil or thickener upon the
bulk properties and the behaviour of the grease.
RL2 and RLS2 are commercially produced by Shell
and they contain some additives. SB-M and SRL
are also available as commercial greases, and are
made by Kyodo Yushi. Both greases contain
additives for which details are not available, while
SB-M is produced specifically as a low-noise grease.
Traction measurements were carried out at an
entrainment speed of 0.02m/s and for several loads
from 10N to 50N. The Hertzian pressure varied
between 0.6GPa and 1.1GPa and the contact
diameter between 170m and 290m. The slide/roll
ratio was varied between zero and 0.6. All tests were
performed at ambient temperature.


4. RESULTS AND DISCUSSION

4.1 Dielectric measurements
The dependence of the real and imaginary
parts of the dielectric constant of SRL grease and its
base oil with frequency are shown in Figure 2.
0
1
2
3
4
100 10000 1000000
Frequency [Hz]
c
'
SRL grease
Base oil

0
0.5
1
1.5
2
100 10000 1000000
Frequency [Hz]
c
"
SRL grease
Base oil

Figure 2. Dielectric permittivity of SRL grease

As explained in the previous section a resistor
has been fitted in series with the measured capacitor,
in order to evaluate the relaxation behaviour of the
PAO oil, which is known to have very low polarity.
This means that the relaxation time extracted form
curves similar to those seen in Figure 2b depends of
Name of Grease Base oil Thickener Weight
percentage (%)
Viscosity (mm
2
/s) at
40 , 100
Worked
Penetration
Grease 1 PAO 12OH-LiSt 12 31, 5.8 236
Grease 2 PAO 12OH-LiSt 12 66, 10 291
Grease 3 PAO 12OH-LiSt 12 411, 41 386
Grease 4 PAO Di-Urea 13.4 31,5.8 280
Grease 5 PAO LiSt 12 19, 4.1 336
Grease 6 PAO LiSt 12 411, 41 339
RL2 HVI160S Di-Urea N/A 107, 12 280
RLS2 PAD1450 Poly-Urea N/A 100, 13 265 - 295
SB-M Synthetic HC Urea 10 - 20 45, 7.7 220
SRL Synthetic Ester LiSt 5 - 15 23, 4.7 250
(a)
(b)

36
the value of this resistor and thus it is not the real
dielectric relaxation time of the oils and grease used
in these tests. Even for polar greases, which
naturally show relaxation behaviour, the relaxation
time could not be measured without the additional
resistor because the maximum is reached at
frequencies outside the measuring range of the
instrument, i.e. 30MHz. Although a quantitative
evaluation of the dielectric relaxation time was
possible, a quantitative comparison of this parameter
between the studied greases and between the grease
and their base oil was performed using this artefact.
For all greases it has been found that they
have higher dielectric constant and longer relaxation
time than the corresponding base oils.
Figure 3 shows the relationship between the
dielectric constant and the relaxation time. From
these results it can be seen that SRL grease, which
consists of a polar base oil and non-polar thickener
has the largest dielectric constant and longest
relaxation time among all ten studied greases.
Greases 1 to 3 which consist of a non-polar base oil
and a polar thickener have relatively high dielectric
constant, while Lithium Grease 5 has the smallest
permittivity and the shortest relaxation time.
1
1.5
2
2.5
3
1.7 2.2 2.7 3.2
Dielectric Constant
R
e
l
a
x
a
t
i
o
n

t
i
m
e
x
1
0
5

[
s
]
LiSt-OH
LiSt
Urea

Figure 3. Dielectric properties of studied greases

Figure 4 shows, comparatively the relaxation
time for polar and non-polar greases. In this Figure,
the name and values inside the boxes indicate the
name of corresponding base oil and its viscosity at
40C and 100C. Also, for better clarity, the scales
in these graphs are different. The distinguishing
feature observed in these graphs is that the greases
made from higher viscosity oil give lager dielectric
constant than that made from lower viscosity oil.
However, the same thing cannot be said about polar
greases. Grease 6 has the largest dielectric constant
between greases of the same type, while grease 1 has
the largest dielectric constant among 12OH-LiSt
greases. The explanation for this behaviour can be
given by the chain lengths of the oil molecules. It is
well known that oils which contain a longer chain in
their chemical structure tend to have high viscosity
and the higher the viscosity of the oil, the higher the
relative permittivity becomes. In the case of non-
polar greases, this mechanism can be applied and the
dielectric constant of bulk grease increases with its
base oils viscosity. On the other hand, in the case of
polar grease, the dipole moment of bulk structure
may become weak when the polar substance is
attached to the long molecular structure. From this
perspective it could be explained why the 12OH-
LiSt grease, formed from a high viscosity base oil,
shows small dielectric constant.

0
1
2
3
2 2.5 3 3
Dielectric Constant
R
e
l
a
x
a
t
i
o
n

T
i
m
e

x
1
0
-
5

[
s
]
.5
12OH-LiSt
LiSt
PAO
411 (40C)
41 (100C)
PAO
66 (40C)
10 (100C)
PAO
31 (40C)
5.8 (100C)
Ester Oil
23 (40C)
4.7 (100C)

(a)
(a)

1.3
1.4
1.5
1.75 1.8 1.85 1.9 1.95
Dielectric Constant
R
e
l
a
x
a
t
i
o
n

T
i
m
e

x
1
0
-
5

[
s
]
LiSt
Urea
PAO
19 (40C)
4.1 (100C)
PAO
411 (40C)
41 (100C)
PAO
31 (40C)
5.8 (100C)
HVI
100 (40C)
11 (100C)
Synthetic Oil
45 (40C)
7.9 (100C)
PAD
100 (40C)
13 (100C)

(b)
(b)
Figure 4. Dielectric properties of polar greases (a)
and non-polar greases (b)

In the low frequency region of the dielectric
measurements, the loss tangent factor,
"
tan
'
c
o =
c
, (4)

37
indicates the electrical conductivity of material [18].
Of all tested greases, only SRL shows good
conductivity, as seen in Figure 5, where the tangent
loss of SRL grease, grease 1, grease 5 and SB-M, are
shown as a function of frequency.

0
1
2
3
1 100 10000 1000000
Frequency [Hz]
t
a
n

o
SRL
SB-M
Grease1
Grease5

Figure 5. Loss tangent factor of four greases

4.2 EHD Traction Measurements
Traction curves of SRL at 10N load are
shown in Figure 6.

0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0 0.2 0.4 0.6 0.8
Slide Roll Ratio
T
r
a
c
t
i
o
n

C
o
e
f
f
i
c
i
e
n
t
1st test
2nd test

Figure 6. Traction curves for SRL grease

In each test, the measurement of the traction
coefficient was carried out twice. After a first test
the rig was left to run for five minutes in pure rolling
conditions and after that, the test was run again in
order to check whether starvation of the contact
occurs, due to the ball pushing the grease to the sides
of the track.
Graphs like those seen in Figure 6 can be
transformed to show the dependence between the
shear stress and shear strain rate. Such dependence
for some of the greases studied in this work is shown
in Figure 7.

0
20
40
60
80
1.E+03 1.E+04 1.E+05 1.E+06
Shear strain rate [s
-1
]
S
h
e
a
r

s
t
r
e
s
s

[
M
P
a
]
SRL
Grease5
Grease6


Figure 7. Shear stress function of shear rate

The traction curves obtained were
subsequently used in an analysis similar to that done
by Evans and Johnson [9], to extract the effective
viscosity q and the Eyring stress of the tested
lubricants. For large shear strain the hyperbolic sine
function can be approximated with an exponential,
and for a negligible elastic term, equation (2)
becomes:
0
t
0 0
0
2
ln ln
| | q
t ~ t + t
|
t
\ .
. (5)
In this relationship, t is the shear stress,
0
t
is the Eyring stress and is the shear rate. In a
logarithmic representation, the slope of the linear
region of the shear stress/shear rate dependence,
seen in Figure 7 for large values of the shear rate,
represents the Eyring stress of the lubricant, as
predicted by Equation (5).
In general, the Eyring stress has been found
to have similar values of that of the base oils, except
four greases, RL2, RLS2, SB-M and SRL, which
showed larger Eyring stress than that of the
corresponding base oils. As seen in Table 1, RL2,
RLS2 and SB-M, have different composition,
although for all of them, the thickener is urea based.
Their base oil is also different from one another:
mineral oil for RL2 and synthetic oils of for RLS2
and SB-M. As shown in Figure 4, these greases
have very similar dielectric relaxation times. The
SRL grease, on the other hand, is very different from
the composition point of view, with a ester base oil
and a lithium soap as thickener. The Eyring stress
for this grease and its base oil can be seen in Figure
8. As this parameter is not directly measured, but it
is calculated as explained above, it is very sensitive

38
to the accuracy of the traction coefficient and film
thickness measurements. For this reason, the tests
have been repeated and the average values, together
with error bars have been shown. It is clear that the
Eyring stress of the grease is larger than that of the
base oil.
0
5
10
15
0.6 0.8 1 1.2
Hertzian pressure [GPa]
E
y
r
i
n
g

s
t
r
e
s
s

[
M
P
a
]
SRL grease
Base oil

Figure 8. Eyring stress for SRL grease and base oil

1
10
100
1000
0.01 0.10 1.00
Entrainment speed [m/s]
F
i
l
m

t
h
i
c
k
n
e
s
s

[
n
m
]
SRL grease
Base oil

Figure 9. Film thickness measurements of SRL
and its base oil

The shear rate involved in the evaluation of
the Eyring stress is determined as the rate between
the sliding speed in the contact and the film
thickness. A comparison of the central film
thickness for the grease and the base oil is shown in
Figure 9. For the low speeds region, where the
traction coefficient tests have been conducted, the
grease has a film thickness about four times larger
than that of the base oil. It is not clear, at this point,
whether the larger value of the Eyring stress for
these greases is a genuine effect, or an experimental
artifact, however, the fact that this tendency
appeared every time the test was repeated, and that
the tests for all greases are done in identical
conditions, suggest indeed, among all greases tested
in this study,that their Eyring stress has larger values
than their base oil.

0
1
2
3
4
0.6 0.8 1 1.2
Hertzian pressure [GPa]
E
f
f
e
c
t
i
v
e

v
i
s
c
o
s
i
t
y

l
o
g
1
0
q

[
P
a

s
]
SRL grease
Base oil

Figure 10. Effective viscosity of SRL grease

0
20
40
60
80
1.E+03 1.E+04 1.E+05 1.E+06
Shear strain rate [s
-1
]
S
h
e
a
r

s
t
r
e
s
s

[
M
P
a
]
Grease4
RL2
RLS2
SB-M

Figure 11. Shear stress for urea-type greases

0.E+00
2.E+07
4.E+07
6.E+07
1.E+03 1.E+04 1.E+05 1.E+06
Shear rate [s
-1
]
S
h
e
a
r

s
t
r
e
s
s

[
P
a
]
Grease1
Grease2
Grease3

Figure 12. Shear stress for 120H-LiSt greases

39
After determining the Eyring stress, from
equation (5) the effective, average viscosity of the
lubricant inside the contact can be extracted. This
was done by fitting equation (5) to the experimental
data. The results for SRL grease are shown in Figure
10. As seen, unlike the Eyring stress, the values of
the effective viscosity of the grease tend to be below
the value of its base oil at high pressure.
From the traction results for all greases, it can
be concluded that urea greases tend to show
relatively high shear stress. An exception to this
rule is SB-M, which shows relatively low shear
stress. This grease, which is produced especially as
low-noise grease, also has the smallest permittivity
among urea greases. Four types of urea greases are
compared in Figure 11. Figure 12 shows the shear
stress function of the shear rate for polar thickener
greases (12OH-LiSt). It can be seen that the shear
stress of grease 3 increases slightly and has lower
value, at high shear rate, compared with the other
greases of the same type. It can be noted that the
dielectric results for these greases showed that
Grease 1 has the largest permittivity, 2.30, while the
dielectric constant of Grease 2 and 3 are 2.15 and
2.06, respectively.
A similar trend was observed for the Li-St
greases shown in Figure 7. Grease 6, which has
slightly larger dielectric permittivity than Grease 5,
also shows larger shear stress.


5. CONCLUSIONS

A dielectric spectroscopy study, coupled with
EHD traction measurements were carried out for a
number of ten greases with various compositions
and properties. The following conclusions have
been drawn form this study.
(1) From the dielectric measurements, it has been
found that greases always have larger dielectric
constant than their base oil and that the base oil
has a stronger influence than the thickener on
the grease dielectric constant. In greases
containing polar components, the dielectric
constant of the grease decreases with increasing
the viscosity of its base oil. In non-polar greases,
on the other hand, the dielectric constant of the
grease increases with the viscosity of its base oil.
(2) The dielectric properties, especially in the low
frequency range, seem to be related to the film
formation in EHD contacts. SRL grease, which
has good conductivity, revealed by the dielectric
measurements, shows poor film formation
ability compared to other greases.
(3) The effective viscosity of grease inside the
EHD contact has been calculated from the
measured value in traction test and
characteristic similar to oil is found in case of
grease. The logarithm of viscosity of greases
varies linearly with pressure, but the degree of
inclination is moderate compared to base oils.
On the other hand, the Eyring stress of greases
rapidly increases and it has been found that for
three of the studied greases the value of this
parameter is greater than that corresponding to
the base oil.
(4) From the traction measurements, it can be
concluded that, in general, urea thickener
greases exhibit large shear stress in their
operation. It has also been found that greases
with higher dielectric constant show larger shear
stress when compared to greases containing the
same type of thickener.


ACKNOWLEDGEMENT

Yuichiro Nagata is grateful to Shell and
Kyodo Yushi for supplying their greases and base
oils and to professor Joichi Sugimura of Kyushu
University, Japan and his colleagues for allocating
their greases from NSK.


REFERENCES

1. Glovnea, R.P. 2010, Transient Phenomena in
Elastohydrodynamic Lubrication, Recent
development in Wear Prevention, Friction and
Lubrications, ed. G. Nikas, Research Signpost Publ.,
pp. 227-262.
2. Cann, P.M.E., 1999, Starved Grease
Lubrication of Rolling Contacts, Tribology
Transactions, Vol:42, pp. 867-873
3. Eriksson P., Wikstroem V. and Larsson R
2000, Grease Passing Through an EHD Contact
Under Pure Rolling Conditions, Proc Instn Mech
Engrs, 214, pp. 309-316.
4. Couronne, I., Mazuyer, D., Vergne, P.,
Truong-Dinh, N., Girodin, D. 2003, Effects of
Grease Composition and Structure on Film
Thickness in Rolling Contact, Trib. Trans.,
vol. 46, 1, pp. 31-36
5. Cann, P.M., Williamson, B.P., Coy, R.C., and
Spikes, H.A., 1992, The Behaviour of Greases in
Elastohydrodynamic Contacts, J. Phys. D: Appl.
Phys, 25.
6. Hirst, W., Moore, A.J., 1979,
Elastohydrodynamic Lubrication at High Pressures-
Non-Newtonian Behaviour, Proc. Roy. Soc. Lond.,
A, 365,pp. 537-565
7. Bair, S., and Winer, WO., 1979, A
Rheological Model for Elastohydrodynamic
Contacts Based on Primary Laboratory Data, J. Lub.
Tech., Vol 101, pp.258-265
8. Johnson, K.L., and Tevaarwerk, J.L., 1977,
Shear Behaviour of Elastohydrodynamic Oil Films,
Proc. R. Soc., Lond., Ser.A., 356, pp.215-236

40

41
9. Evans, C.R., and Johnson, K.L., 1986, The
Rheological Properties of Elastohydrodynamic
Lubricants, Proc Instn Mech Engrs, 200, No C5,
pp. 303-312.
10. Diaconescu, E.N., 2004, Solid-Like Properties
of Molecular Liquids Subjected to EHD Conditions:
Theoretical Investigations, Proc. ASME/STLE Joint
Int. Trib. Conf., 24-27 Oct., Long Beach, USA
11. Furakama, S., and Sumi, T., 1961, A
Dynamic Theory of Piston-Ring Lubrication, Bull.
of JSME, Vol. 4, 16, pp.744-752
12. Hamilton, G.M. and Moore, S.L., 1974,
Measurement of the Oil Film Thickness Between
the Piston Rings and Liner of a Small Diesel
Engine, Proc. Inst. Mech. Engrs., Vol 188, pp. 253-
261
13. Sherrington, I., and Smith, E.H., 1985
Experimental Methods for Measuring the Oil-Film
Thickness Between the Piston-Rings and Cylinder-
Wall of Internal Combustion Engines, Trib. Int.,
vol. 18, 6, pp. 315-320
14. Ibrahim, M., and Cameron, A., 1963, Oil
Film Thickness and Mechanisms of Scuffing in Gear
Teeth, Proc. Inst. Mech. Engrs Lubrication and
Wear Convention, pp. 228-238
15. Vichard, J.P., 1971, Transient Effects in the
Lubrication of Hertzian Contacts, J. Mech. Eng.
Sci., Vol. 13, pp. 173-189
16. Van Leeuwen, H, H. Meijer, H., Schouten,M.,
1987, Elastohydrodynamic Film Thickness and
Temperature Measurements in Dynamically Loaded
Concentrated Contacts: Eccentric Cam-Flat
Follower, Fluid film lubrication-Osborne Reynolds
Centenary, Elsevier, pp. 611-625
17. Nagata, Y., Furtuna, M.D., Bell, C.A., and
Glovnea, R.P., 2008, Evaluation of Electric
Permittivity of Lubricating Oils in EHD
Conditions, Proc. iCAT2008, Dec. 2008,
Singapore, p.155-158
18. Aliotta, F., Fontanella, M.E., Galli, G., Lanza,
M., Migliardo, P. and Salvato, G., 1983, Low-
Frequency Dielectric Investigations in Polymer-like
Lecithin Gels, J. Phys. Chem., 97, pp.733-736.


ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 42-45
Juozas PADGURSKAS
1
e-mail: Juozas.Padgurskas@lzuu.lt
Raimondas KREIVAITIS
1
e-mail: raimondaskreivaitis@gmail.com
Arturas KUPINSKAS
1
e-mail: Arturas.Kupcinskas@lzuu.lt
Raimundas RUKUIA
1
e-mail: raimundas.rukuiza@lzuu.lt
Vytenis JANKAUSKAS
1
e-mail: vytenis.jankauskas@lzuu.lt
Igoris PROSYEVAS
2
e-mail: igorpros@mail.ru
1
Department of Mechanical Engineering,
Lithuanian University of Agriculture - Kaunas,
LITHUANIA
2
Institute of Physical Electronics
Kaunas University of Technology Kaunas
LITHUANIA
INFLUENCE OF NANOPARTICLES ON
LUBRICITY OF BASE MINERAL OIL
The tribological properties of mineral oil SAE 10 modified with
metallic nanoparticles were investigated. The tribological tests
were performed using four-ball test rig. Friction and wear reduction
properties were measured. The positive influence of nanoparticles
on lubricity of mineral oil was observed. The best result was
obtained when using the copper nanoparticles for a single metal
nanosuspension (0.25 % Cu). Using double metal nanoparticles the
best result show iron copper nanosuspension (0.125 % Fe + 0.125
% Cu). The use of those suspensions was most efficient for pure
base mineral oil. The cobalt nanosuspension does not show a
significant increase in lubricity. It was observed that
nanosuspensions stabilize and decrease the friction during the tests.
Keywords: nanoparticles, friction wear, lubricity, mineral oil
1. INTRODUCTION
Presently there is an increasing interest in the
use of nanoparticles. A lot of scientists worldwide
perform many tests simulating the possible
applications of nanoparticles. One of the interesting
nanoparticles applications is to make
nanosuspensions in lubricants. It is observed that
metal or even non-metal nanoparticles have a
positive influence on lubrication properties.
Nanoparticles suspensions in lubricants can
significantly reduce friction and wear of friction
surfaces [1-5].
Such factors as the material, size, shape and
mechanical properties of nanoparticles, their
concentration in the suspension have important
influence on tribological properties of friction pairs
[1,2].
Usually the lubricants are modified by the
nanoparticles of metal or metal oxide (Cu, CuO,
TiO
2
, ZrO
2
, ZnO and so on) and some non-metal
materials (SiO
2
, diamond, inorganic fullerene,
graphite) [1-7]. It is highly evident that only a small
amount of nanoparticles is required to obtain the
positive tribological effect: 0.01 % graphite [6],
<1 % diamond [1], 1 % TiO
2
[1], 5 % IF-WS
2
[4],
0.5...2 % CuO, ZnO, ZrO
2
[3], 0.2 % SiO
2
[7].
Usually, the main research purpose is to estimate
optimal concentration [3,7].
In most cases the commonly manufactured
nanoparticles are simply added to the lubricants [3,
4,6,7]. However there is increasing interest on
synthesis nanosuspensions by the sintering of
nanoparticles directly in the lubricants [8]. The aim
of our study is to estimate the tribological properties
of some nanosuspensions directly sintered in the
base mineral oil.
2. EXPERIMENT
The synthesis of Fe, Cu, Co and Fe+Cu,
Fe+Co, Co+Cu nano-particles is performed by the
CEWLS method when the converse emulsion of
water in lubricant synthesis is used. For that reason
is prepared 100 ml mineral oil SAE10 emulsion with
0.2 ml H
2
O which includes dissolved sulphates of
certain metals, such as FeSO
4
, CuSO
4
, CoSO
4
and
0.5 g cetyltrimenthylammonium bromide (CTAB).
This mixture is poured into 100 ml of mineral oil
SAE10 with 10 ml hydrazine emulsion. All volume
was mixed intensively during 5 minutes. This
mixture with formed according nano-particles was
used for tribological investigations. The
concentrations of nanoparticles in suspensions using
43
single metal were 0.25 %, using double metal
nanoparticles 0.125 % of each metal in compound.
Four-ball type tribotester was used for wear
and friction tests. The balls of 12.7 mm diameter
were made of 100Cr6 bearing steel. The testing
procedure was adapted from the standard
DIN 51 350 [9].
The test oil sample of 22 cm
3
was poured into
the sample chamber, fully submerging the stationary
balls. Under the applied load of 150 N, and rotation
speed of 1420 rpm, the machine was run for 1 hour.
Prior to each experiment, all the appropriate parts of
the machine, i.e. bottom and upper ball holders, oil
vessel and the test balls were washed in an ultrasonic
bath with hydrocarbon solvents, and then dried.
The diameters of the circular wear tracks
(wear scars) on three stationary balls were measured
with an optical microscope. For each run the scar
measurements were reported as an average of the
Wear Scar Diameter (WSD) of the three balls in
millimetres. Friction between the balls is represented
by torque.
3. RESULTS AND DISCUSSIONS
The results of performed tests are presented
in Figure 1. It shows that all used suspensions of
nanoparticles decrease the wear and friction
comparing to pure mineral base oil SAE 10.
However the positive influence was not equal for all
suspensions.
0,0
0,2
0,4
0,6
0,8
SAE 10 SAE 10 +
Fe
SAE 10 +
Cu
SAE 10 +
Co
SAE 10 +
FeCu
SAE 10 +
FeCo
SAE 10 +
CoCu
Test Oil (Composition)
W
S
D
,

m
m
0
10
20
30
40
50
60
70
A
v
e
r
a
g
e

t
o
r
q
u
e
,

m
N
m



.
Wear Average torque
Figure 1. Wear and friction behaviour of base oil and metal nanosuspensions
The best result was obtained using copper
nanoparticles when using single metal
nanosuspension and the worst one when using cobalt
nanoparticles.
Lubrication with copper nanosuspension
reduces the wear 1.87 times and the friction 1.54
times comparing to base oil. Wear reduction was
less important when using iron nanosuspensions in
comparison to copper. However the friction
reduction was virtually the same for both those
nanosuspensions. The wear and friction reduction of
cobalt nanosuspension was almost negligible.
It is likely that wear and friction reduction
properties of copper nanosuspension are related to
its mechanical properties. The soft and ductile
material can easily get between contacting surfaces
and cover them making good protection against the
wear.
Estimation of wear and friction behaviour of
double metal nanoparticles displays that the best
result of all tested nanosuspensions shows iron
copper nanosuspension. Lubrication with iron
copper nanosuspension reduces the wear almost two
times, and the friction 1.76 times. Such a great
increase in lubricity can be due to combination of
two metals with different mechanical properties and
of course different interaction mechanisms.
Despite that cobalt does not reduce the wear
significantly in single metal nanosuspension tests, in
combination with iron and copper it has much better
lubricity. The suspension of cobaltcopper
nanoparticles reduces the wear and friction
respectively 1.78 and 1.75 times comparing to base
oil.
Friction torque variation graphs during the
entire test (Figures 2 and 3) were used for friction
analysis. Figure 2 presents the friction torque when
lubricating with single metal nanosuspension.
44
20
40
60
80
100
0 1000 2000 3000 4000
Test duration, s
T
o
r
q
u
e
,

m
N
m




,
SAE 10
SAE 10 + Cu
SAE 10 + Co
SAE 10 + Fe
Figure 2. Friction torque variation during the test using the lubrication with
pure base oil and single metal nanosuspensions
20
40
60
80
100
0 1000 2000 3000 4000
Test duration, s
T
o
r
q
u
e
,

m
N
m


,
SAE 10
SAE 10 + FeCo
SAE 10 + CoCu
SAE 10 + FeCu
Figure 3. Friction torque variation during the test using the lubrication with
pure base mineral oil and double metal nanosuspensions
Usually friction torque increases during the
tests when using the pure base oil. Regardless some
fluctuations it was also the case in our
investigations. Using the nanosuspensions the
fluctuations almost disappear. Moreover, the
friction value has decreasing trend during the test
when using the nanosuspensions of single copper
and iron.
45
Slightly different results were obtained when
using double metal nanosuspensions (Figure 3).
Friction torque values are stabile for those
suspensions, but we have no friction decrease
tendency for FeCo nanosuspension. It was almost no
difference between the friction behaviour of CoCu
and FeCu nanosuspensions. The reason for that
could be much more important influence of copper
on tribological properties of double metal
nanosuspensions.
4. CONCLUSIONS
Metallic nanoparticles suspensions obtained
by direct sintering in the lubricant are efficient for
wear and frictions reduction. Moreover all used
nanosuspensions help to increase friction stability.
Copper and iron copper nanoparticles
suspensions presented the best lubrication
properties, Co nanosuspensions the worst.
ACKNOWLEDGEMENT
This research was funded by Lithuanian State
Science and Studies Foundation grant B-34/2008
"NanoTribo-Suspensions".
REFERENCES
1. Wu Y. Y., Tsui W. C., Liu T. C., 2007,
Experimental Analysis of Tribological Properties of
Lubricating Oils with Nanoparticle Additives,
Wear, 262, p. 819 825.
2. Hernndez Battez A., Gonzlez R.,
Felgueroso D., and other, 2007, Wear Prevention
Behavior of Nanoparticles Suspension under
Extreme Pressure Condition, Wear, 263, p. 1568
1574.
3. Hernndez Battez A., Gonzlez R.,
Felgueroso D., and other, 2008, CuO, ZrO
2
and
ZnO Nanoparticles as Antiwear Additive in Oil
Lubricants, Wear, 265, p. 422 428.
4. Rapoport L., Leshchinsky V., Lapsker I., and
other, 2003, Tribological Properties of WS
2
Nanoparticles under Mixed Lubrication, Wear, 255,
p. 785 793.
5. Rapoport L., Nepomnyashchy O., Lapsker I.,
and other, 2005, Behavior of Fullerene like WS
2
Nanoparticles under Severe Contact Conditions,
Wear, 295, p. 703 707.
6. Huang H. D., Tu J. P., Gan L. P., Li C. Z.,
2006, An Investigation on Tribological Properties
of Graphite Nanosheets as Oil Additives, Wear,
261, p. 140 144.
7. Peng D. X., Kang Y., Hwang R. M., and other,
2009, Tribological Properties of Diamond and SiO
2
Nanoparticles Added in Paraffin, Tribology
International, 42, p. 911 917.
8. Padgurskas J., Rukuia R., Kreivaitis R., and
other, 2009, Tribological Properties of Mineral Oils
Modified with Metallic Nano-Particles,
Proceedings of 5
th
International Scientific
Conference BALTTRIB 2009, Lithuanian
University of Agriculture, Kaunas, p. 69-76.
9. DIN 51350-3, Testing of lubricants - Testing in
the four-ball tester- Part 3: Determination of wearing
characteristics of liquid lubricants, 1977 (In
Germany).
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 46-51
Alexandru V. RADULESCU
1
e-mail: varrav2000@yahoo.com
Irina RADULESCU
2
e-mail: irena_sandu@yahoo.com
1
Department of Machine Elements and Tribology,
University POLITEHNICA Bucharest,
ROMANIA
2
Mechanical Engineering and Research Institute
S. C. ICTCM S.A. Bucharest,
ROMANIA
INFLUENCE OF THE RHEOMETER GEOMETRY
ON THE RHEOLOGICAL PROPERTIES OF
INDUSTRIAL LUBRICANTS
The rheological properties of two transmission lubricants (75W90
and 75W140), in fresh and used state, were investigated using shear
viscosity rheological measurements. It was found that the lubricants
do not exhibit a yield stress and that, above a critical shear rate, they
exhibit shear-thinning behavior, well described by the Cross model.
The rheological measurements were performed on a Brookfield
viscometer CAP2000+ equipped with four cone-and-plate
geometry, for a range of temperature between 20 70C, using the
viscometer Peltier system. The experiments have shown that only
two geometries are appropriate for testing the transmission
lubricants. Another important conclusion refers to the influence of
the wear degree of the lubricant on the rheological parameter of the
model.
Keywords: viscosity, lubricants, shear-thinning, wear
1. INTRODUCTION
Rheological modeling of lubricants has
always been a subject of great importance when
working with oil from different fields of interest.
The need for predicting the rheological behavior of
the lubricants when experiencing conditions outside
the available measuring range for the equipment
designed in accordance with API specifications [1,2]
has always been present.
Transmission lubricants behave in a non-
Newtonian way. They are shear-rate dependent and
normally termed as shear-thinning lubricants.
Rheological measurements are normally performed
in the laboratory within a given range of shear rates.
Based on these measured values the models should
be able to predict the shear-dependent behavior of
the lubricants outside the measured interval of shear
rates.
The rheological properties of two
transmission lubricants (75W90 and 75W140), in
fresh and used state, were investigated using shear
viscosity rheological measurements. It was found
that the lubricants do not exhibit a yield stress and
that, above a critical shear rate, they exhibit shear-
thinning behaviour, well described by the Cross
model. Supplementary tests have been made,
regarding the thermal behaviour of the lubricants.
Three thermal models have been assumed, in order
to determine the variation of the viscosity with the
temperature, for lubricants in fresh and used state.
2. THEORY
The rheological model based on the Cross
equation is one of the most popular in use today. It
can be found on virtually every research rheometer
software packages and it can be used to extract some
meaningful numbers from the full viscosity versus
shear rate profile (Fig. 1).
The mathematical expression of the the Cross
model is described by the Eq. (1), [3]:
, )
0
m
1

q q
q = q +
+
, (1)
where is the shear rate and is the viscosity at any
given shear rate .
The other parameters involved in eq. (1) are:
-
0
is the zero shear viscosity and represents
the magnitude of the viscosity at the lower
Newtonian plateau. It is a critical material property
and can prove valuable in making assessments of
suspension and emulsion stability, estimating of
comparative polymer molecular weight and tracking
changes due to process or formulation variables etc.
-

is the infinite shear viscosity. This

signifies how the product is likely to behave in very
high shear processing situations such as blade, knife
and roller coating.
- is known as the Cross time constant and has
dimensions of time. The reciprocal, 1/ , gives a
47
critical shear rate that proves a useful indicator of
the onset shear rate for shear thinning.
- m is a dimensionless rate constant (Cross rate
constant) indicating the degree of dependence of the
viscosity on shear rate in the shear-thinning region.
A value of zero for m indicates Newtonian behavior
with m tending to unity for increasingly shear
thinning behavior.
Figure 1. Graphical expression of Cross model
Concerning the thermal model assumed for
the variation of the rheological parameters of the
studied lubricants, three possibilities have been
considered, [4]:
- Jarchov and Theissen model:
50 t
B
95 t
50
e

+
q = q , (2)
where: q lubricant viscosity; q
50
lubricant
viscosity at 50
0
C; B non-dimensional parameter;
t temperature.
- Cameron model:
b
95 t
Ke
+
q = , (3)
where: q lubricant viscosity; K viscosity
parameter; b temperature parameter; t
temperature.
- Reynolds model:
, ) m t 50
50
e

q = q , (4)
where: q lubricant viscosity; q
50
lubricant
viscosity at 50
0
C; m temperature parameter; t
temperature.
In order to obtain the main values of the
characteristic parameters specific for Cross model
(Eq. 1) and all three thermal models (Eqs. (2), (3)
and (4)), the experimental data are numerically
treated, using the regression analysis method, [5].
3. EXPERIMENTAL SET-UP
The rheological measurements were
performed on a Brookfield viscometer CAP2000+
equipped with four cone-and-plate geometry and
using a Peltier system for controlling the
temperature. The CAP 2000+ Series Viscometers are
medium to high shear rate instruments with Cone
Plate geometry and integrated temperature control of
the test sample material, [6]. A typical view of the
viscometer is presented in Figure 2, with all the four
cone and plate geometries.
Figure 2. Geometry of Brookfield viscometer
48
Concerning the technical parameters of the
viscometer, rotational speed selection ranges from 5
to 1000 RPM. Viscosity measurement ranges
depend upon the cone spindle and the rotational
speed (shear rate). Viscosity is selectively displayed
in units of centipoise (cP), poise (P), or Pascal
seconds (Pa.s). Temperature control of sample is
possible between either 5C (or 15C below
ambient, whichever is higher) and 75C or 50C and
235C depending on viscometer model. The
viscometer uses a CAPCALC32 software for
complete control and data analysis. The geometry of
testing cones and the viscosity range are presented in
Table 1.
The lubricants used for testing are two
transmission lubricants (75W90 and 75W140), in
fresh and used state (2000 km), with physical and
chemical properties presented in Table 2, [7]. These
are 100% synthetic extreme pressure lubricants,
characterized by an efficient anti wear protection,
with a better resistance at high temperature and a
longer life time. The lubricants are specially
designed for racing vehicle gearboxes, synchronized
or not synchronized gearboxes, gearbox/differential,
transfer gearboxes and hypod differentials.
Table 1. Geometry and viscosity range of testing cones
Cone number Cone radius, mm Cone angle, degree Viscosity range, Pa.s
3 9.53 0.45 0.083 ... 1.87
5 9.53 1.8 0.333 ... 7.50
6 7.02 1.8 0.833 ... 18.7
8 15.11 3 0.312 ... 3.12
Table 2. Physical and chemical properties of tested lubricants, [7]
Lubricant
Parameter
GEAR 300
75W-90
GEAR Competition
75W-140
Density at 15C (59F) ASTM D1298 900 kg/m
3
906 kg/m
3
Viscosity at 40C (104F) ASTM D445 72.6 mm/s 170 mm/s
Viscosity at 100C (212F) ASTM D445 15.2 mm/s 24.7 mm/s
Viscosity index VIE ASTM D2270 222 178
Flash point ASTM D92 200C 212C
Pour point ASTM D97 -60C -36C
4. RESULTS
The first stage of the experiment was focused
on the influence of the cone and plate geometry on
the measured rheological properties. Figures 3 and 4
show the characteristic rheograms for 75W-90 and
75W-140 oils, in fresh state, with a detail for low
values of the shear rate (0 ... 3500 s
-1
). In the case of
the used oils, the rheograms have the same shape,
similar to those of the fresh oils; that is the reason why
they are not presented in this paper.
Figure 3. Rheograms for 75W-90 oil, in fresh state
By analyzing Figures 3 and 4 it can be
observed that only the cones number 3 and 8 offer
consistent measurements, with low dispersion of the
experimental values. For the viscosity range of the
two oils, cone number 5 is not appropriate for the
experimental tests, due to the large dispersion of the
values.
49
Figure 4. Rheograms for 75W-140 oil, in fresh state
The cone number 8 is characterized by a low
shear rate range (0 . 2000 s
-1
), while the cone
number 3 has a larger shear rate range, between 0 s
-1
and 13300 s
-1
. The tests with this geometry offer
results which can characterize the behavior of the
lubricant for a vaster interest domain.
The experimental results obtained with cones
number 3 and 8 have been treated with the
regression analysis method, according to the Cross
model (Eq. 1), in order to obtain the variation of the
viscosity with the shear rate. The parameters of the
Cross model are presented in Table 3, and the
comparison between the experimental results
obtained with cone number 3 and the theoretical
model is shown in Figures 5 and 6.
It can be observed significant differences
between cone 3 and 8, caused be the extended field
of the shear rate values. Another important
conclusion refers to the influence of the wear degree
of the lubricant on the rheological parameter of the
Cross model. For the 75W90 oil, there are almost
no differences between fresh and used lubricant
(Fig. 5) while the oil 75W140 presents important
changes in used state comparative to fresh state (Fig.
6). The same observation can be made for the two
oils regarding the variation of the viscosity with
temperature.
The experimental results are presented in
Figures 7 and 8 and Table 4 shows the values of the
rheological parameters of the studied lubricants.
Table 3. Parameters of the Cross model for the transmission lubricants
Cross model
Type of oil Wear degree Cone number

q , Pas
0
q , Pas , s m
3 0.153 0.229 7.59510
-4
0.065
fresh
8 0.068 0.275 6.02810
-3
0.147
3 0.118 0.307 3.05710
-3
0.175
75W90
used
8 0.061 0.263 2.04510
-3
0.111
3 0.257 0.628 1.72510
-3
0.104
fresh
8 0.213 0.601 1.21210
-3
0.037
3 0.220 0.574 1.75310
-3
0.095
75W140
used
8 0.151 0.529 7.36210
-4
0.149
75W90 - comparison between fresh and used oil
0.15
0.17
0.19
0.21
0.23
0 2000 4000 6000 8000 10000 12000 14000
Shear rate, 1/s
V
i
s
c
o
s
i
t
y
,

P
a
.
s
fresh - exp. used - exp. fresh - th. used - th.
Figure 5. Comparison between the experimental results and the theoretical model for 75W-90 oil (cone no. 3)
50
75W140 - comparison between fresh and used oil
0.35
0.37
0.39
0.41
0.43
0.45
0 2000 4000 6000 8000 10000 12000 14000
Shear rate, 1/s
V
i
s
c
o
s
i
t
y
,

P
a
.
s
fresh - exp. used - exp. fresh - th. used - th.
Figure 6. Comparison between the experimental results and the theoretical model for 75W-140 oil (cone no. 3)
Table 4. Variation of the rheological parameters of lubricants with temperature
Jarchov and Theissen
model
Cameron model Reynolds model
Parameter
Lubricant
q
50
,
Pas
B
Corr.
coeff.
K, Pas b,
0
C
Corr.
coeff.
q
50
,
Pas
m,
0
C
-1
Corr.
coeff.
Fresh 0. 0446 5. 419 0.9981 1.97910
-4
785.75 0.9981 0. 0435 -0.0473 0.9996
75W90
Used 0. 0446 5. 517 0.9976 1.77610
-4
799.97 0.9976 0. 0432 -0.0478 0.9995
Fresh 0. 1043 5. 232 0.9986 5.57210
-4
758.58 0.9986 0. 1021 -0.0454 0.9995
75W140
Used 0. 0919 5. 520 0.9994 3.67810
-4
800.46 0.9994 0. 0899 -0.0477 0.9989
Figure 7. Variation of the viscosity with temperature for 75W90 oil
Figure 8. Variation of the viscosity with temperature for 75W140 oil
51
5. CONCLUSIONS
The experimental results were found to be well
described by the Cross model, except for the
viscometer geometries number 5 and 6. It can be
observed also significant differences between cone 3
and 8, caused be the extended field of the shear rate
values. Another important conclusion refers to the
influence of the wear degree of the lubricant on the
rheological parameter of the model, including the
variation with temperature.
REFERENCES
1. Qemada, D., 1998, Rheological Modeling of
Complex Fluids I. The Concept of Effective Volume
Fraction Revisited, Eur. Phys. J., Appl. Phys. 1, pp.
119127.
2. *** , 1990, Specification for Materials and
Testing for Well Cements, 5th edn., API Spec., vol.
10, American Petroleum Institute, Dallas, TX, USA
3. Cross, M.M., 1965, Rheology of Non-
Newtonian Fluids: A New Flow Equation for
Pseudoplastic Systems, Journal of Colloid Science,
Vol. 20, No. 5, p. 417-437.
4. Balan, C. (compiled author), 2000, The
Rheology of Lubricating Greases, ELGI,
Amsterdam, 160 p.
5. Crocker, D.C., 1983, How to Use Regression
Analysis in Quality Control, American Society for
Quality Control, Vol. IX, 243 p.
6. *** Oil catalog MOTUL France,
http://www.motul.fr/fichestechniques/Transmission.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 52-57
Vlad Flaviu ZEGREAN
email: zegrean@fim.usv.ro
Emanuel DIACONESCU
Department of Applied Mechanics,
University Stefan cel Mare of Suceava,
ROMANIA
MEASUREMENT OF LUBRICANT OIL
MICROVISCOSITY BASED ON RESONANT
FREQUENCY SHIFT OF AFM CANTILEVER
Experimental investigations on microviscosity of T90 lubricant oil
were conducted using an atomic force microscope. The resonant
frequency of the cantilever beams was measured in air, in pure
water and in sample oil. Based on the resonant frequency shift the
viscosity of the lubricant was calculated using the formula deducted
by Papi [1] for uncalibrated cantilevers. The results obtained are in
good agreement with those of Ionescu [2], measured with a
Rheotest2 rheometer, for the same lubricant.
Keywords: microviscosity, atomic force microscopy, resonant
frequency shift
1. INTRODUCTION
Since its invention by Binning at al. in 1986
[3], the atomic force microscope (AFM) gained
ever-increasing importance in nanotechnology and
biotechnology. The AFM is now widely used to
obtain high-resolution surface topography images, to
measure intermolecular forces or to characterize the
mechanical properties of polymers. The capability
of the AFM to take measurements with the
cantilever and sample submerged in a liquid enhance
the image resolution due to reduced capillary forces.
There are two operating modes for atomic
force microscopy:
- Contact mode, when the cantilever tip is in
permanent contact with the sample. Using
this mode, the friction force between the tip
and the sample can be measured. This
imaging mode dramatically reduces the
lifetime of the tip.
- Tapping mode, when the tip is kept above the
sample by the feedback loop. In this
operating mode, the cantilever is driven at
resonant frequency, near the surface of the
probe by acoustic, magnetic or Brownian
methods. The amplitude and phase of tip
oscillation are then used to extract
information about the sample topography.
The frequency of cantilever vibration
immersed in a fluid strongly depends on the fluid
rheological properties. The drag forces acting on
cantilever are directly related to kinematic viscosity
and fluid density. Thus the viscosity of the fluid can
be determined from the resonant frequency of the
cantilever.
2. THEORETICAL CONSIDERATIONS
To our best knowledge, existing research did
not provide a simple and general relationship
between fluid viscosity and cantilever resonant
frequency. The attempt to find a relationship to
describe the dependence of fluid viscosity on
cantilever resonant frequency can be classified on
two broadly approaches.
Viscous model proposed by Sader [4] and
experimentally validated by Chon at al. [5] accounts
all the geometrical parameters of the cantilever.
Sader assumed that the beam cross section is
uniform over the entire length, the length of the
beam greatly exceeds its width, b, the beam is an
anisotropic linearly elastic solid and internal
frictions effects are negligible, the amplitude of
vibration is far smaller than any length scale of the
beam geometry. The expression that correlates the
normalized Reynolds number, Re , on frequency
response of the cantilever is:
2
vac,1
b
Re
4
e
=
q
, (1)
where is the fluid density,
vac,1
e is the
fundamental radial resonant frequency of the beam
in vacuum and q is the fluid viscosity.
Classical model that draws a heuristic
analogy with the dynamic motion of a sphere trough
viscous fluid was approached by Chen [6], Oden [7],
and Ahmed [8].
The calculation of viscosity using this
approaches rely on accurate values of cantilever
thickness, coating thickness, mass, density and
53
elastic modulus. The calibration procedure of the
cantilever is time consuming, presents the risk of
damaging the cantilever, or has to be reinitiated
when viscosity changes.
To overcome these difficulties, Papi [9]
proposed a method for determining the absolute
value of fluid viscosity by accounting for the
cantilever resonant frequency measured in air or
vacuum, in the sample solution and in a liquid of
known viscosity as a standard.
Based on the classical model, Papi [1] finds
the following mathematical expression to correlate
the sample viscosity,
s
q , to resonant frequencies:
2
2 2
s s
s H O
H O H O
Y
Y
e
q = q
e
, (2)
where
s
e ,
2
H O
e are the resonant frequencies of the
cantilever submerged in sample solution,
respectively in pure water (used as a standard
liquid), and
s
Y ,
2
H O
Y are two dimensionless
parameters yielding from:
2
2
2
2
2 2
0 s,H O
s,H O
2
s,H O
Y
( | |
e e
( | = |
|
e (
\ .
. (3)
The parameter | is a combination of
cantilever beam geometry and structure parameters
and its value can be set as unity for the great
majority of common commercial cantilevers.
3. EXPERIMENTAL PROCEDURE
To measure the resonant frequency of the
cantilever, a Nanonics Imaging Multiview 1000
AFM, depicted in Figure 1, was used. The
cantilever is acoustically driven in a range of
frequencies set by user and the resonant frequency is
detected by the optical device of the AFM.
The Nanonics NWS software allows tracing
the sweep curve, setting the oscillation amplitude of
the cantilever and setting the input gains for the
signal.
Figure 1. Nanonics Imaging Multiview 1000 AFM and Academia optical microscope
In order to submerge the cantilever, a Park
Scientific Instruments Universal SPM liquid cell
was modified to fit the Nanonics sample and probe
mount, Figure 2. A small liquid tank was attached
to a sample mount, Figure 2(a). A transparent thin
flat glass window with a plastic collar was fitted to
the probe mount, Figure 2(b). The probe mount is
magnetically connected to the AFM head.
A spring clasp for standard silicon nitride
probes, Figure 2(b), was firmly attached under the
glass window. The radius of the glass window is
smaller than the interior radius of the tank and it
submerges partially in the liquid along with the
cantilever chip, Figure 2(a). Two mirrors are used to
direct the laser beam from the source on the surface
of the cantilever and the reflected beam to the photo-
sensitive detector, Figure 2(b).
54
(a)
(b)
Figure 2. (a) Modified PSI liquid cell mounted on the Nanonics AFM head; (b) Detail on modified probe mount
Figure 3. Rectangular PSI cantilever
Probe mount
Spring clasp
Liquid tank
Sample mount
Nanonics
AFM head
Spring clasp
Plastic
collar
Glass
window
Probe
mount
Cantilever chip
55
A rectangular shaped Park Scientific
Instruments MicroMash cantilever was used in this
experiment, Figure 3. The laser spot is reflected by
the first mirror on the far edge of the rectangular
cantilever for a maximum deflection angle. The
rectangular cantilever was chosen to reduce the
damping effect on the oscillation when the cantilever
is immersed in the viscous lubricant oil.
This image was captured with a CCD camera
mounted on an Academia optical microscope, Figure
1.
4. EXPERIMENTAL RESULTS
The first frequency sweep curve was traced in
air with the liquid tank empty, Figure 4. The
frequency range was set from 0 to 20 kHz and the
measurements were made in 500 points across this
range. The values displayed are averaged values of
three consecutive measurements in each point. The
maximum amplitude corresponds to the resonant
frequency of the cantilever. The gain set for this
measurement was 0.08. The maximum amplitude of
0.86751 V was reached at a frequency of 11.9038
kHz.
The liquid tank was then filled with pure
water at 20
o
C, measured by a thermocouple and a
new frequency sweep curve, Figure 5, was traced
after the laser was realigned. The realignment is
necessary because refractive index of water modify
the reflection angle of the laser beam. Results were
averaged on ten consecutive measurements for a
precise result. The frequency range was limited to 0
8 kHz because the resonant frequency is lower in a
liquid than in air. The input gain was increased to
0.1 and the maximum amplitude of 1.2133 V at
6.6262 kHz resonant frequency.
After the tank was drained and dried, T90
lubricant oil at 30
o
C was poured into it and the laser
was realigned. Because the lubricant is less
transparent to laser than pure water the input gain
was increased to 3.2. The frequency range was
gradually reduced down to 0 3 kHz. Maximum
amplitude of 3.08057 V was reached at 1.9697 kHz
resonant frequency, Figure 6.
The resonant frequencies experimentally
obtained in air, pure water and T90 lubricant oil are
used in Papis formula (2). Considering the water
viscosity
2
H O
0.00113 Pa s q = , the absolute value
of lubricant kinematic viscosity is
T90
0.265804 q =
Pa s . The result is in good agreement with the
values obtained by Ionescu, [2], on a Rheotest 2
rheometer on T90 lubricant oil in a range of
temperatures from 17
o
C to 100
o
C.
Figure 4. Frequency sweep curve of cantilever in air
56
Figure 5. Frequency sweep curve of cantilever in pure water
Figure 6. Frequency sweep curve of cantilever in T90 lubricant oil
Table 1. Experimental viscosity of T90 measured with Rheotest 2, [2]
Temperature
(
o
C)
Viscosity
(Pa.s)
17 0.4807853
30 0.2726842
40 0.1506939
50 0.0896987
60 0.0574072
70 0.0358795
80 0.0215277
90 0.0143518
100 0.0107638
57
5. CONCLUSIONS
A Nanonics MultiView 1000 AFM from
Laboratory of Micro and Nanotribology at the
University of Suceava was used to determine the
resonant frequency of a rectangular PSI MicroMash
cantilever. To enable cantilever submersion, a Park
Scientific Instruments Universal SPM liquid cell
was modified to fit the Nanonics head.
The microviscosity of T90 transmission
lubricant oil was measured at 30
o
C, by means of
resonant frequency shift of an AFM cantilever in
three different mediums: air, pure water and oil.
The fundamental resonant frequency
measured in air is 11.9038 kHz, decrease to 6.6262
kHz when measured in pure water and reached
1.9697 kHz resonant frequency if submerged in
sample lubricant oil.
The obtained results were applied in Papis
formula for viscosity determination by means of
uncalibrated atomic force microscopy cantilevers.
The absolute value of viscosity,
T90
0.265804 q =
Pa s , is in good agreement with the results obtained
by Ionescu with a Rheotest 2 rheometer.
REFERENCES
1. Papi, M., Arcovito, G., De Spirito, M.,
Vassalli, M., Tiribilli, B., 2006, Fluid Viscosity
Determination by Means of Uncalibrated Atomic
Force Microscopy Cantilevers, Applied Physics
Letters, 88, 194102.
2. Ionescu M., 2004, A System for Viscosity
Measurement in Variable Conditions of Temperature
and Pressure, MOCM 10, Vol.1, 205 208.
3. Binning G., Quate C.F., Gerber Ch., 1986,
Atomic Force Microscope, Phys. Rev. Lett. 56,
930 933.
4. Sader, J.E., 1998, Frequency Response of
Cantilever Beams Immersed in Viscous Fluids with
Applications to the Atomic Force Microscope,
Journal of Applied Physics, 84, 64 76.
5. Chon, J.W.M., Mulvaney, P., Sader, J.E.,
2000, Experimental Validation of Theoretical
Models for the Frequency Response of Atomic
Force Microscope Cantilever Beams Immersed in
Fluids, Journal of Applied Physics, 87, 3978
3988.
6. Chen, G.Y., Warmack, R.J., Thundat, T.,
Allison, D.P., Huang, A., 1994, Resonance
Response of Scanning Force Microscopy
Cantilevers, Rev. Sci. Instrum., 65(8), 2532 2537.
7. Oden, P.I., Chen, G.Y., Steele, R.A.,
Warmack, R.J., Thundat, T., 1996, Viscous Drag
Measurements Utilizing Microfabricated
Cantilevers, Applied Physics Letters, 68, 3814
3816.
8. Ahmed, N., Nino, D.F., Moy, V.T., 2001,
Measurement of Solution Viscosity by Atomic
Force Microscopy, Rev. Sci. Instrum., 72(6), 2731
2734.
9. Papi, M., Maulucci, G., Arcovito, G., Paoletti,
P., Vassalli, M., De Spirito, M., 2008, Detection
of Microviscosity by Using Uncalibrated Atomic
Force Microscopy Cantilevers, Applied Physics
Letters, 93, 124102.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 58-64
M.C. CORNECI
1,2
e-mail: magdalena-carla.corneci@insa-lyon.fr
A.-M. TRUNFIO-SFARGHIU
1
F. DEKKICHE
3,4
Y. BERTHIER
1
M.-H. MEURISSE
1
J.-P. RIEU
3
1
Laboratoire de Mcanique des Contacts et des
Structures, INSA-Lyon, CNRS UMR5259,
F69621 Villeurbanne Cedex, FRANCE
2
Universit Technique Gh. Asachi, Facult de
Mcanique, 700050, Iasi, ROUMANIE
3
Laboratoire de Physique de la Matire
Condense et Nanostructures, Universit Claude
Bernard Lyon 1, CNRS UMR5586, F69622
Villeurbanne Cedex, FRANCE
4
Dpartement de Chimie, Facult de Sciences
exactes. Universit Mentouri Constantine (25000),
ALGERIE
INFLUENCE OF LUBRICANT
PHYSICOCHEMICAL PROPERTIES ON THE
TRIBOLOGICAL OPERATION OF FLUID PHASE
PHOSPHOLIPID BIOMIMETIC SURFACES
Phospholipid bilayers appear to play a key role in joint lubrication
in controlling and reducing frictional forces between biological
surfaces. We have investigated the mechanical and tribological
properties of Dioleoyl phosphatidylcholine (DOPC) bilayers
prepared by the micelle and vesicle method in different solutions
(ultrapure water and Tris buffer pH 7.2 with or without 150 mM
NaCl). Friction forces are measured using a homemade
biotribometer. Mechanical resistance to indentation is measured by
AFM and lipid bilayer degradation is controlled in-situ during
friction testing using fluorescence microscopy. This study confirms
that mechanical stability under shear or normal load is essential to
obtain low and constant friction coefficients. The major result is
that the Tris buffer pH 7.2 improves mechanical and tribological
stability of the studied bilayers. In ultrapure water, bilayers obtained
by the micelle method are not resistant and spontaneously adsorb to
the other contacting surface. Bilayers prepared by the vesicle
method show slightly better lubricant properties than those prepared
by the micelle method. Additional salt (150 mM NaCl) has existing
but secondary effects on the mechanical and tribological properties
of the bilayers.
Keywords: supported phospholipid bilayers, biolubrication, atomic
force microscopy, friction coeficient, nanomechanics
1. INTRODUCTION
Phospholipids, together with proteins, are the
major components of biological membranes and
play vital roles in many biological processes. In
particular, phospholipid layers are found in the
synovial fluid and appear to play a key role in joint
lubrication [1-4]. Other physiological lubricating
sites where lipids are claimed to have a beneficiary
lubricating function include pleura, pericardium, the
ocular surface and the gut where sliding occurs
during gastric motility [5].
Supported phospholipid bilayers (SPB)
composed of phospholipids adsorbed to a planar
solid support are widely used as models to
investigate the properties of these membranes and
associated processes such as molecular recognition,
enzymatic catalysis, cell adhesion and membrane
fusion [6-8]. SPB are also important for a number of
applications including biosensors design, solid
surfaces and biomaterials biofunctionalization,
protein crystallization and DNA immobilization
[7,9,10].
The physical and chemical properties of
biological membranes are of critical importance to
understand specific membrane function. The
stability of SPB is a major concern for the use of
such layers in these various applications. The
atomic force microscopy (AFM) provides an
important way to measure this stability and to
observe the nature of bilayer defects. There have
been several reported studies concerning the effect
of various physicochemical parameters on the SPB
integrity by AFM: pH dependence [11], deposition
pressure of the outer lipid layer [12] or temperature
[13]. AFM has become a major tool to measure
forces between surfaces and colloidal particles [14-
15]. It has also been used to study the interaction
between bilayers and silicon nitride or silicon oxide
tips [16-24].
Besides imaging, force spectroscopy allows
us to obtain valuable experimental information about
the interaction forces and mechanical behavior of the
studied systems with nanometric and nanonewton
resolution through the force-distance curves. Jumps
have been described on the approaching curve on
59
lipid bilayers, this breakthrough being interpreted as
the penetration of the AFM tip through the lipid
bilayer [25]. The force at which this jump in the
force plot occurs is the maximum force the bilayer is
able to withstand before breaking. Thus, a
quantitative measurement of the force at which the
jump occurs can shed light on basic information
concerning cell membrane nanomechanics as well as
interaction forces between neighboring lipid
molecules in the membrane. Therefore, the force
value at which this jump takes place is closely
related to membrane stability [26].
Recently, we observed a clear correlation
between membrane stability (probed by AFM force
curves) and the tribological properties of lipid
bilayers. Our homemade biotribometer allows the
simultaneous measurement of friction coefficient
and the visualization of surface degradation by ligth
and fluorescence microscopy [27]. We have shown
that it is possible to reduce the friction coefficient of
model surfaces by nearly 2 orders of magnitude, to a
very reproducible value =0.002 when both surfaces
are covered by a DPPC (Dipalmitoyl
phosphatidylcholine) bilayer (solid phase). With
only one bilayer in the contact region, the friction
coefficient was much higher than with two bilayers
and it increased during prolonged friction while the
bilayer degraded as evaluated by fluorescence
microscopy. When two DOPC bilayers were
deposited, the friction coefficient was higher than
with DPPC, it increased during prolonged friction
while the bilayer degraded to an extend visibly
depending on the bilayer preparation method [28].
In order to better understand the molecular
mechanisms responsible for the lubricating ability of
SPB it is essential to vary physico-chemical
parameters (temperature and phase of the lipid
bilayer, ions, pH, viscosity of the buffer) and to
measure the structure and the mechanical properties
of the contact (lipid packing, water layers, resistance
of the bilayers to friction or normal load, lipid
mobility and mobility of the fluid around the
bilayer). Our ambition for this study is more
restricted: we aim to measure whether pH, ions and
bilayer method of preparation are changing the
mechanical and tribological properties of SPB.
Resistance to nano-indentation is measured with an
AFM (force spectroscopy mode) while friction and
bilayer degradation under shear are measured with
our biotribometer. The important conclusion of this
study is that DOPC bilayers in an unbuffered
solution (ultrapure water pH 5) are intrinsically less
resistant and lubricant than those in Tris buffer pH
7.2.
2. MATERIALS AND METHODS
2.1 Materials
All chemicals were of analytical grade and
were used without further treatment. 1,2-dioleoyl-sn-
glycero-3-phosphatidylcholine (DOPC) and 1-
palmitoyl-2-{6-[(7-nitro-2-1, 3-benzoxadiazol-4-yl)
amino ] hexanoyl } - sn glycero -3-phosphocholine
(NBD-PC) were purchased from Avanti Polar
Lipids, and used without further purification. NBD-
PC whose ends are fluorescent in blue light was used
to visualize the bilayer homogeneity by fluorescent
microscopy. The buffer beneath the bilayers was 15
mM TrisHCl buffer pH 7.2 (Sigma Aldrich)
containing or not 150 mM NaCl, prepared in
ultrapure water (MilliQ, 18.2 M cm resistivity) and
filtered with a PES membrane 0.20 m before use.
The non-ionic sugar-based surfactant n-dodecyl- -
maltoside (DDM) was purchased from Sigma
Aldrich.
2.2 Preparation of SPB
We used a 8 mm radius convex soft HEMA
lens (Corneal Industrie, Annecy, France) and a flat
borosilicate glass plate as the surfaces on which lipid
bilayers were deposited for tribological
measurements. When swollen in saline solution (150
Mm NaCl, actual pH 7.2), the HEMA lens contains
25% water (wt %) and has mechanical and
physicochemical properties similar to those of
articular cartilage26. Borosilicate glass was also
used for AFM nano-indentation experiments. Glass
surfaces were sonicated twice for 20 min at 60C in
aqueous detergent MicroSon (Fisher-Bioblock,
France) and once for 20 min at 60C in ultrapure
water, then rinsed copiously with ultrapure water
immediately before bilayer deposition. HEMA was
gently cleaned by hand and rinsed copiously with
ultrapure water. SPB were prepared using both
vesicle fusion [29] and micelle [30, 31] methods.
DOPC was solubilized to 1-20 mg/ml depending on
the employed method together with 1% NBD-PC
(wt%) in chloroform/ethanol (9/1, v/v). An
appropriate aliquot was poured in a glass tube and
the solvent was evaporated under a stream of
nitrogen. The resulting lipid film was then kept
under high vacuum overnight to ensure the absence
of organic solvent traces.
To obtain supported phospholipid bilayers
from vesicles (SPBv), we used the following
classical protocol. Multilamellar vesicles (MLV)
were obtained by hydrating the dry lipid film in 15
mM Tris buffer pH 7.2 at room temperature to a
final concentration of 20 mg/ml. MLV were
vortexed for 10 min, frozen for 5 min in liquid
nitrogen and then thawed for 10 min in a water bath,
the whole procedure being repeated six times and in
between each cycle MLV were vortexed for 1 min.
The vesicles were then extruded using a
miniextruder (Avanti Polar Lipids). Samples were
successively subjected to 19 passages through 0.4
and 0.2 m pore diameter polycarbonate membranes
(Avanti Polar Lipids) respectively. The resulting
unilamellar vesicles (LUV) were diluted ten times in
15 mM TRIS buffer pH 7.2 and stored at 4C under
60
nitrogen. Glass and HEMA surfaces were incubated
for 1 hour with the LUV solution diluted ten times,
to which 2 mM of Ca
++
ions were added to stimulate
vesicles fusion and bursting on the surfaces. The
lipidic surplus was then eliminated by rinsing.
Supported phospholipid bilayers were also
produced from mixed micellar solutions (SPBm)
[30, 31]. The dry lipid film was solubilized in
micelles using DDM surfactant and ultrapure water
at 0.114 g/l lipid solution (lipid/DDM = 1/6, wt/wt)
and co-adsorbed on the glass and HEMA surfaces
for 5 minutes in 2 mM of Ca
++
. DDM was
eliminated by slow rinsing with ultrapure water at 3
ml/min for 90 min. A second adsorption from a less
concentrated solution (0.0114 g/l) was sometimes
performed. SPBv or SPBm were conserved in
ultrapure water or 15mM Tris buffer pH 7.2 and
used within a day. 150 mM NaCl was eventually
added to Tris buffer for experiments.
2.3 Atomic force spectroscopy
Measurements were carried out with a
commercial AFM (NanoScope III, Veeco
Instruments, Santa Barbara, CA) equipped with a
liquid J-scanner. Force plots were acquired using V-
shaped Si3N4 tips NP (Veeco) and OMCL
TR400PSA (Olympus, Japan) with a nominal spring
constant of K=0.06-0.12 N/m and K=0.08 N/m
respectively. Individual spring constants were
calibrated using the thermal noise method with a
MFP-3D Asylum Stand Alone AFM. Tip radii of
curvature R were measured by imaging a silicon
grating (TGT1, NT-MDT, Zelenograd, Moscow)
and individual radii were found to be 20-40 nm.
Thousands of approach-retraction cycles were
performed at several locations of the lipidic bilayer
and the cantilever deflection was recorded versus the
position of the Z-piezo of the AFM. These data can
be converted into forcedistance curves where the
force F was calculated from the measured cantilever
deflection z as F=K z, were K is the cantilever
spring constant and subtracting the cantilever
deflection from the height position to obtain the
distance. The tip-sample approaching velocity was
set for all force curves at 400 nms1 so that the
effect of the velocity on the breakthrough force
could be totally neglected. Jump distances,
breakthrough and adhesion forces (Figure 1) were
automatically measured using our own C++ code.
2.4 Tribological measurements
A homemade biotribometer permitting in situ
visualization of the contact was used to measure the
frictional forces between a compliant soft HEMA
lens and a flat borosilicate glass plate, each surface
being covered with one DOPC bilayer as previously
described. An upright epifluorescence microscope
(Leica DMLM) equipped with a fluorescence
camera (Leica DC350F) was used to view the
contact through the opposing glass body. This
observation was performed in situ during friction
and under white and blue light to visualize the
centering of the contact area and the bilayer integrity
respectively. An eddy current position sensor
measured the deformation of the flexible blades
holding the tank, and permitted calculating the
tangential force. An average normal load of 0.3 MPa
was imposed, resulting in a contact area diameter of
about 2 mm independent of the bilayer type. The
friction coefficient was defined as the ratio
between the tangential force (once the surfaces slide
against each other) and the normal load. Several
series of friction tests were performed, each lasting
50 min (about 150 back and forth cycles). Mean and
min-max (for the error bar) values of both initial and
final friction coefficient (i.e., just after the beginning
and after 50 min of friction) were calculated.
3. RESULTS
3.1 Nano-mechanical properties of SPBm
We first present experimental results on the
nano-mechanical resistance to indentation of DOPC
bilayers prepared by micelle method (SPBm). Two
typical AFM approaching-retracting (AR) curves are
displayed in Figure 1 in raw data (i.e., deflexion vs.
Z piezo displacement). The AR curve 1 exhibits a
breakthrough feature in the approaching curve
occurring at a deflexion of about 10 nm
corresponding to a breakthrough force FB ~0.88 nN
(black arrow).
Figure 1. Typical deflection-distance curves
recorded on DOPC SPBm with one incubation. The
cantilever spring constant value is K=0.08 N/m.
Legend: A and R refers as approaching and
retracting curves respectively, numbers as
successive deflection-distance cycles. Cycles 1
exhibit a breakthrough feature with jump of about
4.6 nm in the approaching curve at a force level of
about 0.88 nN and large adhesion peak of about 9.4
nN in the retracting curves (see the enlarged region
in the inset). Cycle 2 on the other hand does not
display either jump or adhesion.
61
As previously reported [26,28,32,33] the lipid
bilayer is unable to withstand the force exerted by
the tip and the breakthrough feature corresponds to
the penetration of the bilayer by the apex of the tip.
The jump distance is about 4.6nm. The
corresponding retracting curves show a large
adhesion force (F
Ad
~ 10 nN). On the other hand,
when the tip does not penetrate the bilayer (curve 2
in Figure 1), the retracting curve does not present
any adhesion.
We have recorded thousands of AR curves up
to a maximal load of 20 nN in water and Tris buffer
pH 7.2 with or without 150 mM NaCl. In ultrapure
water, SPBm (one or two incubations) were
puncturated in 100% of the cases. However, one can
note a clear difference between the two conditions:
the distribution of breakthrough forces is peaked at
about 0.3 nN in the case of one incubation (Figure
2A) while breakthrough forces are higher with a
second histogram peak around 2 nN in the case of
two incubations (Figure 2D). The jump distance and
adhesion force histograms are similar for each
condition (Figures 2B-C, E-F). They are broad and
range between 2 and 10 nm, and between 0 and 20
nN respectively. They are slightly depending on the
sample or on the tip used (each plain bar style
corresponds to a different experiment in Figures 2-
3). We never observed on glass two successive
jumps in the force plots.
When using 15 mM Tris pH 7.2 buffered
solution, the force curves are drastically changing
and very few penetrations are observed either with
one (Figure 2G) or two incubations (not shown). The
penetration frequency is nearly zero when 150 mM
NaCl is added to the Tris buffer with one (Figure
2H) or two incubations (not shown).
Nano-mechanical properties of SPBv. DOPC
bilayers prepared by the vesicle method (SPBv) do
not show any clear dependence on the buffer type
(Figure 3A-F) contrary to SPBm. Nearly 50% of the
AR cycles present no breakthrough feature and the
other are penetrated by the tip under a mean force of
about 10 nN in water or in a Tris buffer pH 7.2
(Figure 3A-D). The associated jump distance
distribution is much more peaked around 3 nm
(Figure 3B-E) than the corresponding distribution
for SPBm (Figure 2B-E). This mean jump distance
corresponds therefore to the thickness of a single
DOPC bilayer34. Another change due to the method
of preparation of bilayers concerns the adhesion
force distribution. When there is penetration of the
bilayer, the mean adhesion value is centered around
3~5 nN for SPBv (Figure 3C, F) instead of 5~10 nN
for SPBm (Figure 2C, F).
3.2 Friction coefficients and degradation of
DOPC bilayers
We have measured the friction coefficient between
hydrophilic surfaces (a convex lens in soft HEMA
articulated against a flat borosilicate glass plate)
each covered or not with a DOPC bilayer. We have
also investigated the effect of buffer and bilayer
method of preparation. Results are summarized in
Figure 4A. In water, as previously found [28], we
have confirmed that as compared to bare surfaces,
is reduced when surfaces are covered with SPB.
However, after 20 min of friction, the value of the
friction coefficient for bare or SPBm covered
surfaces reached the same very high value =0.165
which may induce glass degradation.
Figure 2. Histograms corresponding to the
breakthrough force (A,D,G,H,), the jump distance
(B,E) and the adhesion force (C,F) measured by
AFM for DOPC bilayers prepared by the micelle
methods in different solutions: (A-C) ultrapure
water, one incubation; (D-F) ultrapure water, two
incubations; (G) Tris buffer pH 7.2, one incubation;
(H) Tris buffer pH 7.2, 150 mM NaCl, one
incubation. Colors correspond to different
experiments with different samples and different tips
In Tris buffer pH 7.2, the situation is
drastically changed: on bare surfaces (with or
62
without 150 mM NaCl) the friction coefficient is
stabilized to a lower value =0.1; when surfaces are
covered with SPB, the friction coefficient is
surprisingly low and stable during prolonged friction
(i.e., =0.035 for SPBm and =0.022 for SPBv)
with little effect if any of salt (150 mM NaCl). This
stability of the friction coefficient value in Tris
buffer is accompanied by very little bilayer
degradation if any at the end of the 50 min friction
period as seen by the fluorescence images of the
bilayer (Figures 4B-E) This stability is a really new
result as we previously obtained a large increase of
and an important degradation in water or in a non-
buffered saline solution both by the micelle and
vesicle method [28].
Figure 3. Histograms corresponding to the
breakthrough force (A, D), the jump distance (B, E)
and the adhesion force (C, F) measured by AFM for
DOPC bilayers prepared by the vesicle method in
different solutions: (A-C) pure water; (D-F) pH 7.2
Tris buffer. Colors correspond to different
experiments with different samples and different tips
4. DISCUSSION
In this report, we have found a strong pH
dependence of the mechanical and tribological
properties of DOPC bilayers prepared by the micelle
or vesicle method.
First, SPBm are easily punctured in water but
not in a Tris buffer pH 7.2. Such a buffer influence
does not hold for SPBv. Secondly, the distribution of
jump distances (penetration lengths) is peaked
around 3-4 nm for SPBv but ranges up to 10 nm for
SPBm. The first value compares well with the
bilayer thickness [34,21] while a distance of 78 nm
agrees with the thickness of two DOPC bilayers. It
strongly suggests that, in water and when using the
micelle method of preparation, a bilayer is also
present on the surface of the AFM tip and that two
bilayers are interacting together in the recorded
force-distance curves. This result is in agreement
with previous observations of spontaneous bilayer
adsorption on an initially bare hydrophilic AFM tip
[21,28] or on the glass sphere of a surface force
apparatus when these surfaces are approached to a
DOPC SPBm. Interestingly, such an adsorption was
never reported for the vesicle method to our best
knowledge.
We never observed two successive jumps in
the force plots on glass. The two bilayers are
therefore punctured simultaneously as found for
surfaces covered by surfactant bilayers [36]. The
fact that the adhesion force in the retracting curves
after a penetration is twice larger for the micelle
method reflects probably that a larger force is
needed to separate two adsorbed bilayers (micelle
case) than a bilayer from a bare surface (vesicle
case).
Figure 4. Effect of the bilayer preparation method
and of the buffer on the tribological behaviour of
DOPC supported bilayers. (A) Average values of
friction coefficient are calculated from at least 2
independent experiments. Gray bars represent the
initial value and white bars the final value after 50
min of friction (except two measurements stopped
after 20 min because of too large friction coefficient
degrading the glass surface). Error bars indicate
minimum and maximum measured values.
Abbreviations: W, water; T, Tris buffer pH 7.2; TS,
Tris buffer pH 7.2 with 150mM NaCl; SPBm and
SPBv. (B)-(E) In situ fluorescence visualisations of
the recorded border of the contact zone showing
some eventual bilayer degradation; (B) is the initial
control image (before starting sliding the surfaces)
of a SPBm; the same type of bilayer in water shows
a strong degradation in water after 20 min of friction
(C) and slight degradation in Tris buffer pH 7.2
within 50min of friction (D); a bilayer prepared with
the vesicle method after 50 min of friction in Tris
buffer pH 7.2 with 150 mM NaCl is not degraded
(E)
63
In water, the lower resistance to tip
indentation and the propension to spontaneously
adsorb to other surfaces during contact for SPBm are
certainly related to the presence of traces of
detergent DDM. First, because these phenomena are
not observed in the absence of detergent (vesicle
method) and as secondly because the number of
incubations increased significantly the mean bilayer
breakthrough force. A possible scenario is that
detergent-lipids interactions in water (but not in Tris
buffer) induce a spontaneous curvature and a
destabilization of the bilayer. DOPC is normally
considered to form particularly stable flat bilayers
but in the presence of cholesterol, its spontaneous
radius of curvature is drastically reduced [37].
Cholesterol can even lead to the formation of
nonbilayer structures [38]. The formation of defects
on DLPE SPB during repeated scanning of the AFM
tip was also reported to be highly pH-dependent and
was explained by the increasing bending energy or
frustration due to the high spontaneous curvature of
DLPE monolayers at low pH11. We believe that
such an effect is likely to occur here because of a pH
dependence of residual DDM-lipids interactions.
In Tris buffer pH 7.2, SPBm seem more
resistant to normal indentation than SPBv (Figures
2G,H, 3D). Actually, the distribution of
breakthrough forces for SPBv is similar to
previously reported values on mica with the same
method of preparation [26,33]. The great resistance
of SPBm at pH 7.2 is on the other hand a surprising
result. However, both kind of SPB are not punctured
by forces equivalent to those used in friction
experiments (i.e, 0.3 MPa in a normal joint which
corresponds to about 1nN in AFM experiments)[28].
This resistance to nano-indentation is again
correlated with a weak if any bilayer degradation
during prolonged friction and with a low and stable
friction coefficient. SPBv have slightly better
lubricant properties at pH 7.2 than SPBm, perhaps
again due to the presence of traces of DDM in the
latter case. The friction coefficient with DOPC
bilayers remains however nearly an order of
magnitude larger than DPPC bilayers in the solid
phase previously measured [28]. This indicates that
fluid bilayers are intrinsically less lubricant than
solid ones even if they resist to shear and normal
stress. Understanding the coupling between lipid
mobility and surrounding buffer mobility, along with
localizing precisely the slip plane seems to be the
key steps toward the understanding of mechanisms
of biolubrification by lipid layers.
5. CONCLUSION
We have examined the role of the buffer (pH,
ions) and the bilayer preparation method on SPB
mechanical resistance and tribological properties.
The method of preparation gives very different
properties in water but not in Tris buffer pH 7.2. In
water, SPBm are especially weak and spontaneously
adsorb to the other contacting surface. In a buffer pH
7.2, bilayers are more resistant to nano-indentation
and more stable to prolonged periods of friction than
those in water. They present also better lubricant
properties. Additional salt (150 mM NaCl) have
existing but secondary effects on the mechanical and
tribological properties of the bilayers.
ACKNOWLEDGEMENT
The authors would like to thank Mr. R.
Bougaran of CORNEAL Industrie which provided
us the manufacturated HEMA lenses, Mrs. Piednoir
from LPMCN Lyon and Mr. C. Godeau, from INSA
Lyon for their helpful participation in this work. The
LPMCN team belongs to CellTiss consortium. F.D.
was supported by a PROFAS fellowship.
REFERENCES
1. Hills, B.A., 1995, Ann. Biomed. Eng., 23, 112
115.
2. Johnston, S.A., 1997, Vet. Clin. North Am.
Small Animal Prac., 27, 699.
3. Schwarz, I.M., Hills, B.A., 1998, Br. J.
Rheumatol., 37, 2126.
4. Hills, B.A., 1989, J. Rheumatol., 16, 8291.
5. Hills, B. A., 2000, Proc. Inst. Mech. Eng. H., J.
Eng. Med., 214, 8394.
6. McConnell, H.M., Watts, T.H., Weis, R.M.,
Brian, A., 1986, Biochim. Biophys. Acta, 864, 95
106.
7. Sackmann, E., 1996, Science, 271, 4348.
8. Richter, R.P., Brat, R., Brisson, A.R., 2006,
Langmuir, 22, 3497505.
9. Mou, J, Czajkowsky, D.M., Zhang, Y., Shao,
Z., 1995, FEBS Lett., 371, 279282.
10. Brisson, A., Bergsma-Schutter, W., Oling, F.,
Lambert, O., Reviakine, I., 1999, J. Cryst.
Growth,196, 456470.
11. Hui, S.W., Viswanathan, R., Zasadzinski, J.A.,
Israelachvili, J.N., 1995, Biophysical Journal,
68,171178.
12. Benz, M., Chen, N., Israelachvili, J., 2004, J
Biomed Mater Res A, 71, 615.
13. Fang, Y., Yang, J., 1997, J. Biochim Biophys
Acta, Biomembranes, 1324, 309319.
14. Butt, H.-J., 1991, Biophys.J., 60, 14381444.
15. Ducker, W.A., Senden, T.J., Pashley, R.M.,
1991, Nature, 35, 239241.
16. Dufrne, Y.F., Boland, T., Schneider J.W.,
Barger, W.R., Lee, G.U., 1998, Faraday Discuss.,
111, 7994.
17. Mueller, H., Butt, H.-J., Bamberg, E., 2000, J.
Phys. Chem., 104, 45524559.
64
18. Schneider, J., Dufrene Y.F., Barger W.R., Lee
G.U., 2000, Biophys. J., 79, 11071118.
19. Schneider, J., Barger, W., Lee, G.U., 2003,
Langmuir, 19, 18991907.
20. Butt, H.-J., Franz, V., 2002, Phys. Rev. E., 66,
031601.
21. Grant, L.M., Tiberg, F., 2002, Biophys. J., 82,
13731385.
22. Loi, S., Sun, G. X., Franz, V., Butt , H.-J.,
2002, Experiment. Phys.Rev. E., 66, 031602.
23. Richter, R.P., Brisson, A., 2003, Langmuir, 19,
16321640.
24. Knnecke, S., Krger, D., Janshoff, A., 2004,
Biophys. J., 86, 15451553.
25. Franz, V., Loi, S., Muller, H., Bamberg, E.,
Butt, H., 2002, J. Colloids and Surfaces B:
Biointerfaces, 23, 191200.
26. Garcia-Manyes, S., Oncins, G., Sanz, F., 2005,
Biophysical Journal, 89, 18121826.
27. Trunfio-Sfarghiu, A.-M., Berthier, Y.,
Meurisse, M.-H., Rieu, J.P., 2007, Tribology
International, 40, 15001515.
28. Trunfio-Sfarghiu, A.-M., Berthier, Y,
Meurisse, M.-H., Rieu, J.P., 2008, Langmuir, 24,
87658771.
29. Schnherr, H, Johnson, J. M., Lenz, P., Frank,
C. W., Boxer, S. G., 2004 Langmuir, 20, 11600-
11606.
30. Tiberg, F., Harwigsson, I., Malmsten, M. Eur.,
2000, Biophys J., 29, 196-203.
31. Vacklin, H.P, Tiberg, F., Thomas, R.K., 2005,
Biochimica et Biophysica Acta, 1668, 1724.
32. Oncins, G., Garcia-Manyes, S., Sanz, F., 2005,
Biophysical Journal, 89, 7373-7379.
33. Pera, I., Stark, R., Kappl, M., Butt, H-J.,
Benfenati, F., 2004, Biophys. J., 87, 24462455.
34. Liu, Y., Nagle, J.F., 2004, Phys. Rev. E, 69,
040901.
35. Leroy, S., Steinberger, A., Cottin-Bizonne, C.,
Trunfio-Sfarghiu, A.-M., Charlaix, E., 2009, Soft
Matter, 5, 24 p.4997-5002.
36. Grant, L.M., Ederth, T., Tiberg, F., 2000,
Langmuir, 16, 22852291.
37. (37). Chen, Z., Rand, R.P., 1997, Biophysical
Journal,73, 267276.
38. Coorssen, J.R., Rand, R.P., 1990, Biochem Cell
Biol., 68 (1), 659.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 65-76
Simon LE FLOCH
1,2
e-mail: simon.lefloch@imag.fr
M.C. CORNECI
2,3
A.-M. TRUNFIO-SFARGHIU
2
M.-H. MEURISSE
2
J.-P. RIEU
4
J. DUHAMEL
2
C. DAYOT
2
F. DANG
2
M. BOUVIER
2
C. GODEAU
2
A. SAULOT
2
Y. BERTHIER
2
1
Universit Joseph Fourier, Ecole doctorale
EDISCE, Laboratory TIMC-IMAG, DynaCell
Team, FRANCE
2
Laboratoire de Mcanique des Contacts et des
Structures, INSA-Lyon, CNRS UMR5259,
F69621 Villeurbanne Cedex, FRANCE
3
Universit Technique Gh. Asachi, Facult de
Mcanique, 700050, Iasi, ROUMANIE
4
Laboratoire de Physique de la Matire
Condense et Nanostructures, Universit Claude
Bernard Lyon 1, CNRS UMR5586, F69622
Villeurbanne Cedex, FRANCE
IMAGERIE MEDICALE POUR EVALUER LES
CONDITIONS DU FONCTIONNEMENT
TRIBOLOGIQUES DES ARTICULATIONS
SYNOVIALES
Le but de ce travail est dvaluer avec prcision les conditions
tribologiques macroscopiques subies par larticulation du genou au
cours de la marche en essayant de considrer les interactions entre
elles. Le contexte plus global est la comprhension du
fonctionnement tribologique de larticulation saine lui permettant de
durer 70 ans. Dans une premire partie, les vitesses relatives
tangentielles entre les surfaces en contact ont t values au cours
de la marche. Ltude a t consacre aussi lvaluation des
conditions gomtriques du contact lorsque le pied subit un effort
de compression de lordre de 300 N (patiente de 29 ans pesant 60
kg ayant subi une mniscectomie).
Les rsultats sur la dforme sont valids qualitativement par des
lments bibliographiques. Ils permettent dmettre une hypothse
quant la capacit du cartilage se dformer de quelques diximes
de mm suivant son paisseur sans que la pression locale de contact
soit importante, permettant une rpartition de la pression trs
efficace.
Ltude a t complte par lvaluation de linfluence des efforts
musculaires sur la raction de contact et sur les dplacements
relatifs des os (patient de 36 ans et de 62 kg (avec une lsion des
mnisques) a subi une compression du membre infrieur). Il est
conclu que laction des muscles augmente normment la pression
moyenne de contact, mais que cette action peut aussi permettre
doptimiser les conditions de contact en dplaant le tibia par
rapport au fmur.
Keywords: articulation synoviales, conditions du contact,
compression de cartilage
1. INTRODUCTION
Larticulation saine du genou est un systme
tribologique performant car elle peut fonctionner
normalement plus de 70 ans. La comprhension de
son fonctionnement tribologique a particulirement
t tudie dans les dernires dcennies car les
maladies osto-articulaires prennent de plus en plus
dampleur et les traitements ne savrent souvent pas
assez efficaces.
De nombreuses tudes ont montr le
fonctionnement complexe des diffrentes
composantes articulaires prises individuellement
(cartilage, synovie et ensemble musculaire) mais le
fonctionnement rel dune articulation dpend non
seulement des proprits individuelles de chaque
composant articulaire mais aussi des interactions et
de larrangement structurel entre tous ces
composants.
Do notre intrt de comprendre le
fonctionnement tribologique de lensemble de
larticulation.
Pour cela des tudes rcentes ont permis de
dfinir un modle exprimental ex-vivo prenant en
compte les caractristiques physico-chimiques et
mcaniques de lensemble in vivo cartilage
synovie.
Ce modle permet de comprendre lchelle
microscopique le fonctionnement tribologique de
linteraction cartilage synovie.
En revanche ce modle comporte des
paramtres quil faut fixer par rapport aux conditions
relles du contact (pressions de contact, aires de
contact, dforme du cartilage, vitesses relatives
tangentielles) qui font lobjet de cette tude.
66
2. ETUDE IN VIVO DU MECANISME
ARTICULAIRE DU GENOU
Ce travail a t ralis avec la collaboration
avec le Laboratoire de Biomcanique et
Modlisation humaine (LBMH, Universit Lyon 1)
et le service dArthroscanner de lHpital Lyon Sud.
L'objectif de ce travail est dvaluer in vivo
les conditions tribologiques macroscopiques
(pressions de contact, vitesses relatives entre les
cartilages articulaires) subies par une articulation du
genou au cours de la marche.
Dans ce domaine, la bibliographie n'apporte
en effet que des informations souvent incompltes,
et manquantes d'inter cohrence.
Pour obtenir ces conditions tribologiques,
nous avons entrepris une tude englobant:
la dtermination de la cinmatique du contact
(vitesses tangentielles relatives),
la mesure de la gomtrie du contact
(dimensions et courbures, conformit des corps
en contact, aires de contact),
la dtermination de la dynamique du contact:
mesures quasi statiques de la dforme des
cartilages articulaires et de la variation de
laire de contact en fonction du chargement
articulaire ;
calculs des efforts musculaires et de la
raction du contact, estimation de la
pression,
lvaluation du rle musculaire dans
l'optimisation des conditions de contact
(diminution de la pression de contact par un
dplacement relatif des os).
3. STRATEGIE
Nous avons valu les conditions
cinmatiques locales, lors de la marche dune
personne saine : partir de donnes issues de la
bibliographie (Kapandji ), et dune base de donnes
sur la marche construite au sein du LBMH, des
simulations ont t effectues en dynamique inverse
afin de calculer la raction de contact .
Par imagerie mdicale (arthroscanner), nous
avons valu les conditions gomtriques locales in
vivo (courbure des corps en contact, aires de
contact) Ces essais ont galement permis une
analyse quasi-statique de la phase dappui de la
marche, dans le but final dvaluer la rpartition de
pression en contact.
Deux patients bnvoles ont particip nos
essais quasi-statiques en arthroscanner. Au cours de
ces essais, nous avons mesur leffort extrieur
appliqu sur le pied de chaque patient. Nous avons
galement pu enregistrer lactivit lectrique des
muscles de l'un des deux patients laide dun
dispositif lectromyographique.
Ces lments ont permis de calculer les
efforts dvelopps par les muscles articulaires du
genou, grce une valuation in vivo des bras de
levier des muscles et lutilisation dun code de
calcul dvelopp au sein du LBMH (thse en cours
dAlice Bonnefoy).
La mesure du dplacement relatif des os a
permis dtudier le rle des muscles pour
loptimisation des conditions tribologiques de
contact.
Une brve description anatomique du genou
est prsente ci-dessous, ainsi qu'un rsum de
rsultats obtenus dans le cadre de ce travail.
4. ANATOMIE DE GENOU
Larticulation du genou est une diarthrose,
c'est--dire une articulation mobile, qui comporte
(Figure 1):
Deux segments osseux, prsentant des
surfaces articulaires. Dans le cas du genou, une des
extrmits osseuses est reprsente par les condyles
fmoraux et lautre extrmit par le plateau tibial;
Deux cartilages articulaires recouvrant les
surfaces osseuses (cartilage fmoral et cartilage
tibial);
La capsule articulaire, fibreuse;
La membrane synoviale, qui tapisse
intrieurement la capsule;
Des ligaments, reliant les deux segments
osseux (pour le genou il y a 4 ligaments : deux
ligaments croiss et deux ligaments latraux);
Des petites structures fibro-cartilagineuses,
les mnisques, qui viennent sinsrer sur les surfaces
articulaires pour assurer une meilleure conformit
des surfaces. Larticulation du genou prsente deux
mnisques (l'un interne et lautre externe) attachs
au plateau tibial par des petits ligaments;
Une poche graisseuse et des bourses sreuses,
sortes de petits coussins hydrauliques constitus
dune enveloppe membranaire contenant un liquide
similaire au liquide synovial.
Enfin, les tendons des muscles qui sinsrent
proximit dune articulation entourent lensemble
de ces structures. Le rle principal des muscles est
de mobiliser larticulation, mais ils ont aussi une
importance pour maintenir la cohsion de
larticulation (coaptation).
De nombreux travaux consacrs l'anatomie
par imagerie mdicale ont permis de dfinir avec
exactitude la gomtrie des extrmits osseuses en
contact. Ainsi, les deux condyles fmoraux forment
des cyclodes avec des rayons maximaux de 38mm
pour le condyle interne et 60 mm pour le condyle
externe (Figure 2.).
En ce qui concerne le plateau tibial, il
prsente deux cavits (glnes) correspondant aux
deux contacts avec les condyles fmoraux. Les deux
glnes ont des rayons denviron 70mm. Elles se
67
distinguent par leur courbure, la glne interne tant
concave et la glne externe convexe (Figure 3).
Cela gnre un contact conforme pour le condyle
interne et un contact non conforme pour le condyle
externe.
5. CINEMATIQUE DU GENOU
5.1. Modle de calcul des vitesses
Pour tudier la cinmatique du genou, nous
avons utilis lhypothse classique de rattachement
ferme des mnisques au plateau tibial. Ainsi, la
cinmatique simplifie du genou, consiste en la
composition de deux mouvements entre le condyle
fmoral et la glne tibiale :
du roulement, influenc par les courbures des
corps en contact,
du glissement correspondant des
translations relatives entre les deux
extrmits osseuses.
Figure 1. Reprsentation anatomique de larticulation du genou
Figure 2. Gomtrie des condyles fmoraux
Figure 3. Gomtrie des glnes tibiales
68
Un modle gomtrique simple a t utilis
pour les condyles. Ils sont reprsents chacun par un
arc de cercle. Le plateau tibial est reprsent par
deux segments de droite (Figure 4). Ce modle, de
type cylindre sur plan, permet de quantifier
rapidement les vitesses tangentielles entre les
surfaces.
Figure 4. Modle gomtrique de condyle pour le
calcul des vitesses
Selon les conventions de signe utilises
(vitesses positives selon les X positifs), la vitesse
relative au point de contact de la surface du cartilage
dun condyle du fmur par rapport la surface du
cartilage du tibia est donne en prenant le tibia
comme solide fixe de rfrence. La vitesse relative
tangentielle au point de contact du fmur par rapport
au tibia est la somme de la vitesse du fmur en ce
point avec la vitesse du point gomtrique de
contact:
fmur / tibia I
V (I) V .R (1)
ou R est le rayon du condyle considr, V
I
la vitesse
du point de contact I, la vitesse angulaire.
Les valeurs de rayons de courbure retenues
correspondent 10 de flexion du genou, et sont
issues de louvrage de rfrence de Kapandji [2]: 55
mm pour le condyle externe et 35mm pour le
condyle interne.
Le dplacement des points de contact en
fonction de langle de flexion, est donn par Li et al.
[4] pour le cas des genoux sains en flexion passive.
La bibliographie dmontre, malgr tout, que les
dplacements des condyles sont modifis suivant
que le pied est charg ou non.
Deux vitesses tangentielles sont calcules aux
deux points de contact . Enfin, une ide de la
raction du contact articulaire du genou est donne
par les calculs de dynamique inverse. Cet effort
nest quune estimation qualitative de leffort de
contact, permettant davoir une ide des conditions
interactionnelles entre la cinmatique et la pression
de contact.
5.2. Rsultats
Les rsultats des essais sur la cinmatique du
genou sont prsents dans la Figure 5.
Les vitesses relatives de roulement entre les
surfaces au niveau du contact du condyle interne
varient cycliquement de 0 mm/sec 200 mm/s au
cours dune marche effectue 5 km/h. Pendant la
phase dappui, la vitesse relative change deux fois de
signe et sa valeur maximale dans cette phase est de
80 mm / s pour le contact interne et de 130 mm / s
pour le contact externe. Pendant la phase
doscillation (faibles pressions de contact), les
vitesses relatives sont plus importantes (interne : 200
mm / s ; externe : 300 mm / s) et changent galement
deux fois du signe.
Figure 5. Rsultats des essais sur la cinmatique du genou
69
Les vitesses de glissement du contact vers
larrire du plateau tibial (25 mm / s au maximum)
ne compensent pas les vitesses relatives induites par
la rotation du fmur par rapport au tibia car les
rayons de courbure des condyles sont assez
importants. Le condyle externe, dont la surface de
contact se dplace plus vers larrire que le condyle
interne, a un rayon de courbure suprieur, ce qui
augmente la vitesse relative tangentielle entre les
surfaces en contact 300 mm/s au maximum.
Ces rsultats sont trs sensibles aux rayons
des condyles retenus. Par contre, lamplitude de la
translation des points de contact vers larrire du
plateau tibial (de 0 60 de flexion), ninfluence
pas significativement le calcul. Cette translation a
une valeur maximale de 10 mm.
6. DINAMIQUE DU GENOU
6.1. Modle de calcul des dformes des cartilages
articulaires (fmur et tibia)
La stratgie consiste valuer la dforme in
vivo du cartilage des deux condyles fmoraux en
faisant une diffrence entre les paisseurs de
cartilage mesures avant et aprs la compression de
larticulation.
Pour obtenir les paisseurs de cartilage in
vivo nous avons utilis le scanner (rayons X) du
CHU de Lyon Sud. La technique de scanner utilise
est nomme arthro scanner . Elle comporte
linjection intra articulaire dun produit base diode
qui permet de faire ressortir sur des images
radiologiques les parties cartilagineuses de
larticulation. Le volume scann par les rayons X est
de 15 x 23 x 20 cm
3
et il permet davoir 44 + 77 +
66 coupes de genou dans les trois plans du repre
anatomique (Figure 6).
Un montage a t ralis pour comprimer le
membre infrieur du sujet lors de la prise dimages
par irradiation X.
Le traitement des deux lots dimages
(charg/non-charg) a consist tout dabord en une
reconstitution des volumes du cartilage du fmur.
Cette reconstruction a t ralise en utilisant le
logiciel AutoCAD (Figure 7), et a permis de reprer
les zones de contact et mesurer les aires de contact
extrieurement charg ou non. Un recalage spatial, a
t ncessaire pour superposer le volume du
cartilage comprim sur le volume du cartilage non
comprim, en faisant concider des points de repre
dfinis sur l'os.
Figure 6. Reconstitution du volume de cartilage en ArthroScanner
Figure 7. Volume du cartilage comprim,
(AutoCAD)
Une carte de la dforme du cartilage du
fmur a t ainsi ralise.
Le calcul de leffort de contact a t fait en
adaptant le modle de dynamique inverse notre
tude. Ainsi, dans le cadre de notre tude, les
acclrations sont nulles (quasi statique), et le
torseur au niveau du pied se limite une force
deux composantes (F impact). Un capteur force
permet de mesurer les deux composantes de cette
force sur le pied (suivant X et Y). Le problme est
donc plan, et ne permet que le mouvement de
flexion extension au niveau du genou. Les quatre
groupes de muscles inter articulaires du genou les
plus importants pour la flexion extension sont
70
Figure 8. Modle de dynamique inverse pour calculer les efforts musculaires et la raction de contact
inclus dans le modle : le quadriceps, le biceps
femoris, le semitendinosus et le gastrocnemius. Les
bras de leviers articulaires sont mesurs in vivo
grce aux images darthroscanner. Ces donnes sont
introduites dans le modle de calcul en dynamique
inverse qui value les efforts musculaires
dvelopps par le membre infrieur. (Figure 8)
Pour le calcul en dynamique inverse, deux
critres physiologiques ont t retenus : on minimise
la raction de contact, ainsi que la norme
quadratique des contraintes des muscles.
Pour vrifier de manire exprimentale si les
muscles sont contracts ou non, un dispositif
lectromyographique a t utilis.
Par contre, ce dispositif permet seulement
dvaluer de manire qualitative la contracture
musculaire.
Deux patients ont particip notre tude :
Une patiente de 29 ans, pesant 60 kg. Elle na
plus de mnisques sur le genou droit, duquel
est effectu le scanner. Le fait quelle nait
plus de mnisques augmente les
dformations. Leffort de compression a t
de 300 N. Nous avons ralis une carte de
dformes du cartilage du fmur, une analyse
daires de contact fmur - tibia (avant et aprs
compression).
Un patient de 36 ans qui pse 62 kg. Il a une
lsion au niveau des mnisques sur le genou
gauche o est effectu le scanner, en
revanche, les mnisques sont en trs grande
partie sains. L'effort de compression a t de
310 N. Pour cette tude, nous avons effectu
un calcul deffort de contact en prenant en
compte laction musculaire.
6.2. Rsultats
Les efforts extrieurs au niveau du pied qui
ont t considrs dans ce travail, avec des vitesses
nulles entre les surfaces articulaires et un angle
presque nul entre le fmur et le tibia correspondent
la fin de la phase dappui de la marche (points
rouges de la Figure 5).
Dans ces conditions, la pression moyenne au
niveau du contact articulaire du genou avec
mnisques, a t value ente 10
5
et 10
6
MPa. On a
constat que cette pression est augmente de 50%
s'il ny a pas de mnisques.
De plus, cette tude nous a permis de montrer
linfluence des muscles sur les conditions de contact
articulaire, cela est schmatis dans la Figure 6.
Ainsi :
Le rle principal des muscles au niveau dune
articulation est de gnrer des couples internes
permettant dassurer le mouvement et/ou lquilibre.
Il a t montr que les efforts externes (poids propre
de 40 N et effort de 330 N au bout du pied)
appliqus au membre infrieur tendent rduire la
flexion au niveau du genou. Dans ce cas, les
flchisseurs dveloppent un effort de lordre de 1200
N pour compenser les efforts externes. Grce un
enregistrement des efforts musculaires au cours de la
compression de la jambe (lectromyogramme), il a
t conclu que le quadriceps est galement actif.
Lactivit musculaire du quadriceps nest cependant
pas ncessaire en terme de couple. Cette activit,
quil nous est impossible de quantifier grce au code
de calcul utilis pour valuer lactivit musculaire
des autres muscles, doit probablement permettre de
stabiliser larticulation. Ainsi, la valeur de 200 N
indique Figure 9 (en opposition 0 N), pour
laction du quadriceps est totalement arbitraire. Plus
leffort du quadriceps est important, plus leffort des
flchisseurs doit tre important pour compenser le
couple interne dvelopp par le quadriceps. Ainsi, si
un effort de 200 N est considr au niveau du
quadriceps, leffort dvelopp par lensemble des
muscles flchisseurs est de 1500 N.
Un rle secondaire des muscles est de
modifier les positions relatives du tibia par rapport
71
au fmur, ce qui provoque une tension du ligament
crois postrieur et une compression de la partie
antrieure des mnisques. Ce mouvement modifie la
rpartition de pression et donc les conditions du
contact.
Pour cette tude nous nous sommes placs
dans les conditions de fonctionnement dfavorables
pour la lubrification articulaire par effets de type
hydrodynamique, qui correspondent la phase situe
entre 15% et 45% du cycle de marche (rectangle
jaune de la Figure 5).
Dans cette phase, on peut considrer que les
pressions de contact atteignent entre 10
5
et 10
6
MPa,
avec des vitesses relatives variant de 0 5 cm/s.
Figure 9. Schma densemble du rle du mcanisme dans les conditions du contact articulaire du genou
(Avec mnisques / sans mnisque)
7. CONCLUSIONS
La bibliographie apporte de manire spare
et de faon incomplte les informations ncessaires
ltude de la tribologie au niveau de larticulation
saine du genou. Deux grands buts ont donc t
poursuivis tout au long de cette tude. Le premier
but tait de mieux spcifier les conditions de contact
telles que les vitesses de frottement entre les
surfaces de contact, les aires de contact, la pression
de contact, la dforme du cartilage et les efforts
musculaires.
Le second but tait de mettre en relation ces
lments qui sont considrs sparment dans la
bibliographie. En effet, les chelles considres
pour comprendre et analyser les diffrents
phnomnes vont du mtre (analyse du mouvement
densemble comme la marche), au m (tude
tribologique du mcanisme de larticulation).
Pour synthtiser les apports majeurs de ce
travail, trois schmas ont t crs.
Le premier schma (Figure 9) prsente le rle
des diffrents lments mcaniques, principalement
le rle des muscles, dans le systme tribologique
constitu des cartilages et des mnisques (les
premiers corps), du liquide synovial (le troisime
corps) et lensemble des muscles, des os et des
ligaments (le mcanisme). Il souligne limportance
du mcanisme de larticulation sur les conditions
tribologiques du contact.
En effet, le premier rle des muscles au
niveau dune articulation est de gnrer des couples
internes permettant dassurer le mouvement et/ou
lquilibre. Ce premier rle influe sur la pression de
contact qui est augmente de manire considrable.
Ainsi, la pression moyenne au niveau du contact
entre les cartilages est value 1,5 MPa. Cette
pression est en grande partie issu de laction des
72
muscles qui compriment larticulation. Si les actions
des muscles ne sont pas considres dans le calcul
de la pression moyenne, cette pression moyenne
nest plus value qu 0,33 MPa. Un rle
secondaire mais trs probable des muscles est de
modifier les positions relatives du tibia par rapport
au fmur, ce qui provoque une tension du ligament
crois postrieur et une compression de la partie
antrieure des mnisques (ces efforts ne sont pas
valus au cours du cette tude). Ce mouvement
modifie la rpartition de pression et donc les
conditions du contact. Dans le cadre de notre tude,
les efforts externes (poids propre de 40 N et effort de
330 N au bout du pied) appliqus au membre
infrieur tendent rduire la flexion au niveau du
genou. Les flchisseurs dveloppent un effort de
lordre de 1200 N pour compenser les efforts
externes. Grce un enregistrement des efforts
musculaires au cours de la compression de la jambe
(grce un lectromyogramme), il est conclu que le
quadriceps est aussi actif. Lactivit musculaire du
quadriceps nest pas ncessaire en terme de couple.
Cette activit, quil nous est impossible dterminer
grce au code de calcul utilis pour valuer lactivit
musculaire des autres muscles, doit probablement
permettre de stabiliser larticulation. Ainsi, la
valeur indique de 200 N (en opposition 0 N), pour
laction du quadriceps est totalement arbitraire. Plus
leffort du quadriceps est important, plus leffort des
flchisseurs doit tre important pour compenser le
couple interne dvelopp par le quadriceps. Ainsi, si
un effort de 200 N est considr au niveau du
quadriceps, leffort dvelopp par lensemble des
muscles flchisseurs est de 1500 N. La pression
moyenne de contact est alors rvalue 2 MPa. La
tension dans le ligament crois postrieur et la
compression de la partie antrieure des mnisques
sont alors moins importantes que lorsque le
quadriceps est relch.
Les efforts extrieurs appliqus au niveau du
pied sont analogues ceux rencontrs la fin de la
phase dappui de la marche. La flexion de la jambe
nest alors pas la mme, modifiant toutes les
conditions tribologiques tudies en statique au
cours de ce travail. Les vitesses relatives
tangentielles exposes sur le schma (Figure 9) sont
celles de la fin de la phase dappui, ne donnant
quune indication des sollicitations en cisaillement
que le gel synovial subit en fin de phase dappui. La
rpartition de la pression de contact est influence
par laction des muscles, mais aussi par les vitesses
tangentielles relatives entre les surfaces en contact.
En effet, ltude sur les vitesses relatives
tangentielles au niveau des points de contact permet
de conclure un rgime de lubrification de type
squeeze alli un rgime de type palier en
fin de la phase dappui, ces rgimes modifiant la
rpartition de pression en contact. Allie la
gnration de pression par le cisaillement et
lcrasement du gel synovial, la dforme du
cartilage influe elle aussi sur la rpartition de
pression. Avec une amplitude de 0,9 mm pour des
efforts extrieurs proches de ceux rencontrs en fin
de phase dappui, la dforme du cartilage ne peut
tre nglige lors de ltude de la rpartition de
pression.
Les muscles prcdemment cits sont activs
par le systme nerveux qui dtient des informations
sur les conditions de contact. Le second schma
(Figure 10) expose les possibles retours
dinformations (jusquau systme nerveux) sur les
conditions tribologiques du contact exposes
prcdemment. Ainsi, la tension des ligaments
articulaires (internes et externes), la tension de
chaque muscle, leffort subit par les mnisques (par
lintermdiaire de la membrane articulaire) et la
pression de contact (par lintermdiaire de la
pression sanguine des vaisseaux de los sous le
cartilage) sont connus par le systme nerveux. Le
systme nerveux peut en retour contrler certains
paramtres (dplacements, efforts, douleur si il y a
des lsions) grce des critres physiologiques quil
se fixe. Ce contrle vient influencer les mesures
effectues et surtout vient influencer les conditions
prcdemment exposes. Nous lavons vu dans
cette tude, les muscles, contrls par le systme
nerveux, peuvent modifier de manire importante les
conditions de contact, pour une mme position
macroscopique enregistre. Pour une mme position
statique (il est srement possible dtendre cette
observation au cas du mouvement), il est ainsi
possible de rduire la pression de contact ou
dannuler les efforts dans les ligaments croiss et la
pression de contact sur la partie antrieure des
mnisques.
Ces informations engranges par lorganisme
peuvent aussi avoir des consquences sur les
proprits mcaniques du cartilage (module
dYoung, coeff. de Poisson, paisseur), mais plus
long terme. Dans la priode embryonnaire la
transformation du tissu cartilagineux (le modle d'os
embryonnaire) en tissu osseux est dtermine par
linvasion des capillaires sanguins dans le tissu
cartilagineux (processus d'ossification). Ce
processus est stopp au niveau articulaire par
l'quilibre entre la pression mcanique exerce au
niveau articulaire (par des mouvements articulaires)
et la pression de perfusion des capillaires sanguins
(exerce dans le processus d'ossification). Cette
zone d'quilibre des pressions marque la zone du
contact entre l'os et le cartilage articulaire. Cette
zone est spcifique pour chaque individu et elle peut
voluer au cours de la vie en fonction de l'intensit
de l'effort mcanique transmit dans l'articulation
(condition physique de chaque individu). Ainsi, le
manque de mouvement diminue l'paisseur du
cartilage articulaire et peut mme produire une
ossification complte de l'articulation dans le cas de
manque de mouvement dans la priode
embryonnaire.
73
Figure 10. Schma des retours dinformations possibles vers le systme nerveux concernant les paramtres du
mcanisme de larticulation. Hypothses sur les consquences mcaniques possibles
De part ses mnisques et les formes
complmentaires des surfaces de contact du fmur et
du tibia, larticulation du genou a une bonne
conformit lchelle du cm. En effet, les
mnisques et les formes complmentaires des os
(surtout pour le condyle interne), permettent davoir
une trs bonne rpartition de la pression. Les
courbures des surfaces de contact sont alors de
lordre de quelques cm (4 8 cm). Ltude du
contact entre les cartilages doit sintresser des
chelles plus petites. A ce niveau, il a t montr au
que le cartilage se dforme de manire importante,
jusqu 0,9 mm de diminution de lpaisseur lors
dune compression de lordre de 1500 N sur
larticulation.
Une hypothse qui peut expliquer ces fortes
dformations calcules est la structure mme du
cartilage qui pourrait se dformer de manire
importante sans dvelopper des pressions
importantes au niveau de sa surface. Cette hypothse
permettrait daccommoder les diffrences de
gomtrie entre les surfaces en opposition jusqu
quelques diximes de mm, toujours dans le but de
rpartir au mieux la pression. Par exemple, au
niveau des bords internes des mnisques, le cartilage
du fmur peut se dformer pour saccommoder la
marche que reprsente le bord du mnisque (marche
de quelques diximes de mm).
A lchelle du m, prise en compte grce au
travail en cours au sein du LaMCoS sur la
comprhension du fonctionnement tribologique
dune articulation saine (Ana-Maria Sfarghiu), la
conformit du contact est assure par la formation de
sacs ou de poches de gel synovial permettant
dassurer une continuit de la pression malgr la
rugosit du cartilage.
Ces trois chelles diffrentes concernant la
conformit du contact permettent une rpartition trs
efficace de la pression de contact, limitant long
terme lusure (Voir Figure 11).
Un autre aspect de cette tude a t
lexploration de techniques permettant de retirer des
informations grce aux images du scanner (ou de
lIRM). Grce ces images, les bras de levier des
muscles, pour une certaine position du membre
infrieur, ont t valus in vivo avec une bonne
prcision. Lamplitude de la dforme maximale du
cartilage a elle aussi t value in vivo : Cette
technique permettra dans les annes venir, avec
lamlioration de la rsolution des images, de
connatre in vivo et sans perturber le contact les
dformations de chaque lment de larticulation.
74
Figure 11. Les trois chelles importantes assurant la conformit du contact pour une articulation saine
ACKNOWLEDGEMENT
Les auteurs remercient vivement Mme
Laurence Chze (LBMH), Monica Cretan, Alice
Bonnefoy, dr. Xavire Rivire et son quipe de
radiologie (Hpital Lyon Sud), Lionel Lafarge
(INSA de Lyon) pour laide prcieux apport cette
tude.
REFERENCES
1. Adriacchi, O. Dyrby, 2005, Interactions
Between Kinematics and Loading During Walking
for the Normal and ACL Deficient Knee, Journal
of Biomechanics, Vol. 38.
2. Adriacchi, O. Dyrby, 2004, Secondary
Motions of the Knee During Weight Bearing and
Non-Weight Bearing Activities, Journal of
Orthopaedic Research, Vol. 22.
3. Ahmed, Burje, 1983, In-vitro Measurement of
Static Pressure Distribution in Synovial Joints--Part
I: Tibial Surface of the Knee, J Biomech Eng..
4. Bassey et al., 1997, Relations Between
Compressive Axial Forces in an Instrumented
Massive Femoral Implant, Ground Reaction Forces,
and Integrated Electromyographs from Vastus
Lateralis During Various Osteogenic Exercises,
Journal of Biomechanics, Vol. 30.
5. Beillas et al., 2004, A New Method to
Investigate In Vivo Knee Behaviour Using a Finite
Element Model of the Lower Limb, Journal of
Biomechanics, Vol. 37.
6. Bergmann et al., 2001, Hip Contact Forces
and Gait Patterns from Routine Activities, Journal
of Biomechanics, Vol. 34.
7. Brewer Robin, 2002, Reconstitution du
fonctionnement du genou naturel et artificiel, DEA
au LaMCoS / CHU Lyon Sud.
75
8. Cheng et al., 2001, The Influence of Inserting
a Fuji Pressure Sensitive Film Between the
Tiobiofemoral Joint of the Knee Prosthesis on the
Actual Contact Characteristics, Clinical
Biomechanics, Vol. 16.
9. Clair David, 2000, Analyse et modlisation des
effets mcaniques dans le processus dusure par
impacts / glissements. Application des contacts de
gomtrie conforme. Thse au LMC (actuel
LaMCoS).
10. Colloud Floren, 2003, Modlisation dynamique
du rameur lors dexercices raliss sur ergomtres
daviron. Implications pour lentranement, thse
dirige par Laurence Chze UCB 1.
11. Doriot Nathalie, 2001, Modlisation dynamique
du membre infrieur pour lestimation des forces
articulaires et musculaires mises en jeu pendant la
phase dappui de la marche, Thse UCBL 1 sous la
tutelle de Laurence Chze.
12. Duda and Taylor et al., 2004, Tibio-Femoral
Loading During Human Gait and Stair Climbing,
Journal of Orthopaedic Research, vol. 28.
13. Fleisig et al., 1998, An Analytical Model of the
Knee for Estimation of Internal Forces During
Exercise, Journal of Biomechanics, vol. 31.
14. Freeman et al., 2005, The movement of the
normal tibio-femoral joint, Journal of
Biomechanics, Vol. 38.
15. Gautier Thomas, 1997, Adaptation du modle
danalyse quantifie de la marche ltude clinique
des IMC, Laboratoire LBMH UCB Lyon 1, 2004.
16. Glitsch et al., The Three-Dimensional
Determination of the Internal Loads in the Lower
Extremity, J. Biomechanics, Vol. 30.
17. Haut Donahue et al., 2002, How the Stiffness
of Menisci Attachments and Meniscal Material
Properties Affect Tibio-Femoral Contact Pressure
Computed Using a Validated Finite Element Model
of the Human Knee Joint, Journal of Biomechanics,
Vol. 36.
18. Hobatho et al, 1998, In Vivo Determination of
Contact Areas and Pressure of the Femorotibial Joint
Using Non-Linear Finite Element Analysis,
Clinical Biomechanics, Vol. 13.
19. http://catalog.nucleusinic.com: site commercial
dimages danatomie.
20. Huiskes et al., 1997, An Inverse Dynamics
Modelling Approach to Determine the Restraining
Function of the Human Knee Ligament Bundles, J.
Biomechanics, Vol. 30.
21. Iwaki et al., 2000, Tibio-Femoral Movement 1:
The Shapes and Relative Movements of the Femur
and Tibia in the Unloaded Cadaver Knee: Studied by
Dissection and MRI, Journal of Bone and Joint
Surgery, Vol. 82B.
22. Johal et al., 2004, Tibio-Femoral Movement in
the Living Knee. A Study of Weight Bearing and
Non-Weight Bearing Knee Kinematics Using
Interventional MRI, Journal of Biomechanics,
Vol. 37.
23. Kapandji, 1965, Physiologie articulaire:
schmas comments de mcanique humaine. 2;
Membre infrieur. Maloine.
24. Komistek et al., 2004, Knee Mechanics : A
Review of Past and Present Techniques to
Determine In Vivo Loads, J. Biomechancis, Vol.
38.
25. Kurosawa, Fukubayashi, Nakajima, 1980,
Load-Bearing Mode of the Knee Joint: Physical
Behaviour of the Knee Joint With or Without
Menisci, Clin Orthop, Jun.
26. Li et al., 2004, In Vivo Tibiofemoral Contact
Analysis Using 3D MRI-Based Knee Models,
Journal of Biomechanics, Vol. 37.
27. Lu et al., 1998, Validation of a Lower Limb
Model With In Vivo Femoral Forces Telemetered
from Two Subjects, Journal of Biomechanics, Vol.
31.
28. Mahfouz et al., In Vivo Determination of the
Normal and Anterior Cruciate Ligament-Deficient
Knee Kinematics, Journal of Biomechanics, 2004
29. Majumbar et al., 2004, A Three Dimensional
MRI Analysis of the Knee Kinematics, Journal of
Orthopaedic Research, Vol. 22.
30. Maquet, Van de Berg, Simonet, 1975,
Femorotibial Weight-Bearing Areas. Experimental
Determination, J Bone Joint Surg Am., Sep.
31. McCarthy et al., 1999, Knee Cartilage
Topography, Thickness, and Contact Areas from
MRI: In-Vivo Calibration and In-Vivo
Measurements, Oseoarthristis and Cartilage, Vol.
7.
32. McPherson et al., 2004, Imaging Knee Motion
Using MRI, RSA/CT and 3D digitalization,
Journal of Biomechanics, Vol. 37.
33. Pandy et al., 1997, A Musculoskeletal Model
of the Knee for Evaluating Ligament Forces During
Isometric Contractions, J. Biomechanics, Vol. 30.
34. Pinskerova et al., 2001, The Shapes and
Relative movements of the Femur and Tibia in the
Unloaded Cadaver Knee: A Study Using MRI as an
Anatomic Tool. Chapter 10. In: Insall, J. Scott, W.N.
(Eds), Surgery of the Knee, Edition Saunders.
35. Ramsey et al., Tibiofemoral contact points
relative to flexion angle measured with MRI,
Clinical Biomechanics, Vol. 17, 2002.
36. Rotat Franois, 2005, Rapport de projet de fin
dtude effectu au sein du dpartement de lINSA
genie mcanique et conception.
37. Scarvell et al., 2004, Comparison of
Kinematics in the Healthy and ACL Injured Knee
Using MRI, Journal of Biomechanics, Vol. 38.
38. Scarvell et al., 2004, Evaluation of a Method
to Map Tibiofemoral Contact Points in the Normal
Knee Using MRI, Journal of Orthopaedic
Research, Vol. 22.
39. Shelburn et al., 2004, Pattern of Anterior
Cruciate Ligament Force in Normal Walking,
Journal of Biomechanics, Vol. 37.
76
40. Shirazi-Adl et al., 1995, Biomechanics of the
Human Knee Joint in Compression: Reconstruction,
Mesh Generation and Finite Element Analysis, The
Knee, Vol. 2.
41. Shrive, OConnor, Goodfellow, 1978 ;
42. Taylor S.J.G. et al., 2001, Forces and
Moments Telemetred from Two Distal Femoral
Replacements During Various Activities, Journal
of Biomechanics, Vol. 34,.
43. Thambyah et al., 2004, Contact Stresses in the
Knee Joint in Deep Flexion, Medical Engineering
and Physics.
44. Thambyah et al., 2005, Estimation of Bone-
on-Bone Contact Forces in the Tibiofemoral Joint
During Walking, The Knee.
45. Walker PS, Erkman MJ., 1975, The Role of
the Menisci in Force Transmission Across the Knee,
Clin Orthop.
46. Walsh et al., 1999, An Improved Method for
Measuring Tibiofemoral Contact Areas in Total
Knee Arthroplasty: A Comprarison of K-Scan
Sensor and Fuji Film, Journal of Biomechanics,
Vol. 32.
47. Wang et Walker, 1974 ;
48. Wo et al., 1991, Tibial meniscal dynamics using
three-dimensional reconstruction of magnetic
resonance images.
49. Woo et al., 1997, Biomechanics of the ACL:
Measurments of In Situ Force in the ACL and Knee
Kinematics, The Knee, vol. 5.
50. You et al., 2001, In Vivo Measurement of 3D
Skeletal Kinematics from Sequences of Biplane
Radiographs : Application to Knee Kinematics,
IEEE Transactions on Medical Imaging.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 77-84
M.C. CORNECI
1,2
e-mail: magdalena-carla.corneci@insa-lyon.fr
A.-M. TRUNFIO-SFARGHIU
1
F. DEKKICHE
3,4
Y. BERTHIER
1
M.-H. MEURISSE
1
J.-P. RIEU
3
M. LAGARDE
5
M. GUICHARDANT
5
1
Laboratoire de Mcanique des Contacts et des
Structures, INSA-Lyon, CNRS UMR5259,
F69621 Villeurbanne Cedex, FRANCE
2
Universit Technique Gh. Asachi, Facult de
Mcanique, 700050, Iasi, ROUMANIE
3
Laboratoire de Physique de la Matire
Condense et Nanostructures, Universit Claude
Bernard Lyon 1, CNRS UMR5586, F69622
Villeurbanne Cedex, FRANCE
4
Dpartement de Chimie, Facult de Sciences
exactes. Universit Mentouri Constantine (25000),
ALGERIE
5
Institut Multidisciplinaire de Biochimie des
Lipides, INSERM / INSA de Lyon, UMR870,
F69621, Villeurbanne, FRANCE
PHOSPHOLIPIDES DANS LE FLUID
SYNOVIAL - INFLUENCE SUR LE
FONCTIONNEMENT TRIBOLOGIQUE DES
ARTICULATIONS SYNOVIALES
PATHOLOGIQUES
Des tudes rcentes ont montr le rle des assemblages lipidiques
associes la structure discontinue du fluide synovial dans les
performances du fonctionnement tribologique dune articulation
saine. Dans le cas des pathologies articulaire, ce fonctionnement
est modifi. Ce travail cherche ainsi identifier linfluence de la
variation en composition lipidique des fluides synoviaux
pathologiques sur le fonctionnement tribologique des articulations
synoviales atteintes de diffrentes pathologies (arthrite, arthrose)
Keywords: biotribology, synovial joints, phospholipids, lipidomic
analysis
1. INTRODUCTION
Les recherches de ces dernires annes ont
propos diffrents types de substances en tant que
responsables (seules ou ayant une action synergique)
pour la lubrification au niveau des articulations
synoviales caractrises par un frottement bas et une
usure rduite. On peut mentionn : lacide
hyaluronique (HA) [1], la lubricine [2-4], le
cartilage, qui attire les ttes hydrophiles des
phospholipides [5] (en utilisant la chromatographie
sur couche mince (CCM), Hills avait conclu que la
phpsphatidyl-choline (PC) est le plus abondant
phospholipid quon trouve sur la surface du cartilage
articulaire en y format des structures empiles [6,7]),
les proteoglycans de type PRG4 [8], les SAPLs [2].
Ces composants sont synthtises au niveau de
fluide synovial par les synoviocites : PGL4 [9],
SAPL [2], HA [10], les chondrocites [11].
En tant que fluide biologique, le fluide
synovial est un ultrafiltrate de plasma, concentre
travers la membrane synoviale.
Lquilibre (lhomostasie) entre la
production des composants de fluide synovial et leur
filtration au niveau de la membrane synoviale assure
le comportement tribologique remarquable des
articulations saines, constituant des systmes ayant
une usure rduite et un frottement bas au niveau du
contact articulaire.
Pourtant, dans le cas des pathologies
articulaires (larthrite, larthrose ainsi que le
remplacement dune articulation non fonctionnelle
par un implant articulaire) on constate une
augmentation du frottement au niveau du contact
articulaire ainsi que lusure des surfaces de cartilage
qui est accompagne par une altration des
caractristiques biochimiques du fluide synovial,
notamment de sa composition lipidique.
Dans le cas de larthrose, lusure des
cartilages a t corrle entre autre avec des
disfonctionnement biologiques dus lactivation de
la phospholipase A2 qui modifie la structure
macromolculaire du fluide synovial en dtruisant
ses assemblages lipidiques, ce qui modifie leur
78
comportement mcanique et par consquent celui de
larticulation synoviale.
Dautres marqueurs des disfonctionnements
biologiques (ex. lucotrine de type LTB4 et
prostaglandine de type 5-HETE dans les cas
dinflammation) ont t identifis dans larticulation
arthrosique.
Mme si les tudes ont montr que les
phospholipides sont les composants majeurs
contribuant la lubrification au niveau de cartilage
articulaire [14,15], il ny a pas beaucoup dtude
sintressant la nature des diffrentes classes de
phospholipides prsents au niveau des articulations
synoviales (lies la surface de cartilage articulaire
ou bien dans la composition du fluide synovial).
Cette tude est donc mene afin danalyser le
contenu en phospholipides prsentes dans diffrentes
clases de fluides synoviaux pathologiques. Des
contraintes dthique mdicale nous ne permettent
pas davoir accs pour ces tudes des chantillons
de fluide synoviaux sains.
Les diffrentes classes de phospholipides ont
t identifies et quantifies en utilisant la
chromatographie gazeuse (GC) en couplage avec la
chromatographie sur couche mince (CCM).
En vingt ans, la pratique mdicale a constat
un accroissement des maladies articulaires de plus
de 30%. Leurs traitements supposent soit
lutilisation de mdicaments soit lintervention
chirurgicale pour mettre en place des prothses
articulaires dont la dure de vie in vivo est au
maximum de 10 ans, alors que la dure de vie dune
articulation saine est denviron 70 ans.
Des tudes rcentes [12] considrent que
cette diffrence peut-tre due aux conditions d'essais
ex vivo, qui gnralement ne respectent pas
suffisamment la ralit biologique complexe. En
effet la plus parte des pathologies articulaires sont
traites en essayant damliorer les constats
cliniques. Si cette dmarche donne des rsultats
assez satisfaisants pour les pathologies simples, les
rsultats satisfaisants sont en nombre limit pour le
cas des pathologies complexes au niveau des
contacts frottants biologiques et cela parce que dans
ces cas il y a plusieurs composants qui doivent tre
considrs.
Lors dune pathologie articulaire, cest tout le
triplet tribologique qui est soumis des
dysfonctionnements biologiques. Si ces
dysfonctionnements sont nos jours de mieux en
mieux diagnostiqus individuellement, ils en restent
encore des difficults exprimentales in vivo qui
rendent assez difficile la prdiction de leurs effets
coupls ainsi que lidentification des lments du
triplet tribologique quils affecteront. Cette situation
fait que souvent un traitement nintervient que sur
un effet intermdiaire sans mme parfois traiter le
bon lment du triplet. Pour remdier cette situation,
il faut donc optimiser les traitements pour cibler
llment du triplet tribologique dans lequel il faut
contrler la libration du bon principe actif. Pour ce
faire, dans cet tude on cherche reproduire ex vivo
les pathologies articulaires afin doptimiser leurs
traitements. Pour cela, on utilise le modle
tribologique ex-vivo raliste [13] qui prend en
compte le comportement biologique complexe dune
articulation (cartilage et fluide synovial) (Figure 1).
Ce modle a permis de mettre en vidence le
rle des assemblages lipidiques, associs la
structure discontinue du fluide synovial (Figure 2),
qui assurent les performances tribologiques du
fonctionnement articulaire sain. De plus, il respecte
les structures molculaires et les interactions entre
les lments du triplet tribologique - cet dire le
mcanisme (le fonctionnement articulaire saine,
simplifi avec une vitesse de glissement trs faible,
permettant dexacerber le rle de 3
me
corps), les 1er
corps (reproduisant ex vivo les caractristiques du
cartilage articulaire sain (hydrogels type HEMA), et
le 3me corps (reproduisant dune manire raliste la
structure et la composition du fluide synovial sain).
Figure 1. Modle ex vivo raliste de la lubrification articulaire [12]
Figure 2. Assemblages lipidiques dans la structure discontinue du fluide synovial [12]
79
Dans le cas dune pathologie articulaire de
type arthrite on a une inflammation du fluide
synovial accompagne dune baisse locale de pH, ce
qui favorisent laugmentation du coefficient de
frottement et des endommagements locale qui
dterminerons lusure du cartilage, la principale
caractristique dune arthrose qui ncessite pour son
traitement la mise en place dune prothse. Vue ces
changements au niveau dune articulation
pathologique on peux distinguer des paramtres des
disfonctionnements pathologiques introduire dans
le modle ex vivo existant: mcaniques,
physicochimiques et biologiques.
Dans ce contexte, lobjectif de ce travail est
dadapter le modle ex vivo existant en y
introduisant les paramtres biologiques,
physicochimiques et mcaniques des pathologies
afin de simuler le fonctionnement pathologique et de
comprendre le bon enchanement cause/consquence
responsable dune pathologie et donc de cibler son
traitement aussi bien mdicamenteux
(pharmacologie articulaire) que prothtiques
(surfaces frottantes des implants articulaires).
Il faudra donc apporter des solutions
doptimisation des traitements pour diffrentes
pathologies articulaires et pour un meilleur
fonctionnement tribologique des articulations
prothses.
2. MATERIEL ET METHODE
Pour arriver proposer des solutions
doptimisation des traitements des pathologies
articulaires il est ncessaire de faire des tudes
biologiques pour dterminer les variations de la
composition lipidique [13] et la destruction
dassemblages lipidiques par des actions
enzymatiques associs aux pathologies articulaires.
Pour cela, la composition lipidique/
biologique relle de ces assemblages lipidiques,
dans le cas de pathologies articulaires ou en
prsence dune prothse ont t dtermines en
utilisant des analyses lipidomiques de fluides
synoviaux caractristiques pour diffrentes
pathologies articulaires (collaboration avec lInstitut
Multidisciplinaire de Biochimie des Lipides, INSA
de Lyon, France) [16].
a. Lipides analyss : phosphatidylcholine (PC),
phosphatidylthanolamine (PE) et phosphatydil-
inozitol + phosphatidil srine (PI + PS)
Extraction lipidique : On a fait la sparation
des lipides en diffrentes classes : phospholipides,
mono- et di-glycrides, cholestrol, acides gras
libres, triglycrides et esters de strol par
chromatographie sur couche mince (CCM) ensuite la
sparation des diffrentes classes de
phospholipides : PE, PC, PI, PS. Ensuite, par trans-
estrification de ces fractions on a fait l analyse de
leur contenu en acides gras par chromatographie
gazeuse couple la spectromtrie de masse.
b. Lipides mdiateurs de linflammation
Linfiltration des leucocytes entrane la
formation de leucotrines et en particulier du
leucotrine B4 (LTB4) form partir de lacide
arachidonique (20:4n-6). Ce dernier est libr par
activation de la phospholipase A2 lors de
linflammation et devient substrat de la 5-
lipoxygnase pour former les leucotrines et lacide
5-hydroxy-eicosattranoque (5-HETE). LTB4 et 5-
HETE ont t extraits du liquide synovial, puis
mesurs par chromatographie liquide haute
performance (HPLC).
Ces deux marqueurs ont t validits dans de
nombreux modles inflammatoires [15-17]. Ils
permettront dans les tudes cliniques de dterminer
si la prothse est bien accepte par le patient. Ils
pourront aussi tre utiliss pour vrifier lefficacit
de mdicaments anti-inflammatoires.
Quatre types de fluides synoviaux
pathologiques ont t analyss :
fluide synovial suite au dclement et
remplacement dun implant articulaire (I) ;
fluide synovial dans le cas dune arthrose
localise (AL) ;
fluide synovial dans le cas dune arthrose
localise accompagne dune infection
(ALI) ;
fluide synovial dans le cas dun patient
manifestant une arthrose gnralise (AG).
c. Prparation des chantillons biologiques (dans
chaque prouvette on a 1 ml fluide synovial)
A lhpital, des prcautions simposent : il
faut utiliser seulement des rcipients en verre, jamais
de plastique ; de plus la solubilisation de la
synovie se fait sur place dans des prouvettes en
verre contenant chaque une : 3 ml thanol, 0.15 ml
deferoxamine (agent hmolytique, 15mol final,
M=656.79g/mol), 0.5 ml butilhydroxytolun - BHT
(antioxydant, 5mmol final, M=220.35g/mol) et
ensuite le transport au laboratoire se fait dans les
plus brefs dlais dans une bote de glace carbonique
20C). Une fois arriv au laboratoire on ferme
sous azote et on stock les chantillons -20C
jusquau moment quand on continue lanalyse
lipidomique, cet dire lextraction et
quantification des lipides;
Extraction lipidique
Les lipides sont extraits [18] laide dun
mlange thanol/chloroforme (3/6, v/v) contenant du
BHT la concentration finale de 5.10-5 M, en
milieu acide (pH = 3) ( laide dacide actique
glaciale). On utilise comme standard interne du
100l PC (10mg/ml) et 50l PE (1mg/ml) ajouts
dans chaque tube.
Les lipides sont extraits selon la mthode de
Bligh et Dyer (1959) par le mlange
80
chlorophorme/mthanol (1/2 ; v/v) en prsence de
BHT (5.10-5M). Aprs fermeture sous azote, vortex
et centrifugation (5 min, 25C, 1800torr/min) le
mlange se spare en deux phases. La phase aqueuse
est rcupre et transfre dans un autre tube,
ensuite vapore sous jet dazote ; une deuxime
extraction est effectue el les phases organiques
obtenues sont runies puis vapores sec et
conserves sous azote -20C. On ajoute de
nouveau 3ml thanol et 6 ml de chloroforme et
lchantillon est de nouveau centrifug ensuite le
surnageant (la phase suprieure) est rcupr et la
phase restante est r extraite 2 fois comme
prcdemment.
Sparation par CCM (en phospholipides
totaux, marqueurs doxydation)
Les extraits lipidiques (vapores sec sous
azote) sont reprises dans un faible volume du
mlange mthanol/chloroforme (1/2, v/v) (250l)
pour tre dposes sur une plaque de silice de
chromatographie en couche mince (CCM). Les
standards de migration sont dposs simultanment
raison de 10l LTB4 et 8 l 9-HODE. La
sparation des lipides est effectue par migration
dans le systme de solvants : n-hxane : diethil
ther : acide actique glaciale (25 :75 :1, v/v/v).
Apres la sparation, et le schage des plaques, la
rvlation des standards de migration se fait par
vaporisation de phosphomolibdate puis chauffage de
la plaque 50C.
La zone de silice correspondant au dpt est
situe la mme hauteur que le standard de
migration plus 1 cm au dessus et au dessous. Les
bandes de silice correspondant aux phospholipides
totaux et aux marqueurs doxydation sont grattes et
rcupres dans des tubes en verre vis. La silice est
rhydrate par un mlange de 2 ml mthanol :
chloroforme (2/1) et ensuite 2 fois avec 2 ml
chloroforme, pour les phospholipides totaux et 2 fois
avec 3 ml mthanol pour les marqueurs doxydation
afin dextraire ces composants de la silice. Les
chantillons sont centrifugs (5min, 25C,
1800torr/min). La phase suprieure est rcupre
pour prparer la sparation CCM des phospholipides
totaux [19] et de lautre cot pour prparer les
chantillons pour HPLC (identification des
marqueurs de stress oxydant). Les phases dintrt
sont rassembles et vapores sec sous azote.
Sparation par CCM (des phospholipides)
Les extraits de phospholipides totaux
(vapores sec sous azote) sont reprises dans un
faible volume du mlange mthanol/chloroforme
(1/2, v/v) (400l) pour tre dposes sur une plaque
de silice de chromatographie en couche mince
(CCM). Les standards de migration sont dposs
simultanment raison de 30l PE (1mg/ml) et 3l
PC (10mg/ml). La sparation des lipides est
effectue par migration dans le systme de solvants :
chloroforme : mthanol : mthyle amine aqueuse
40% (61 :19 :5, v/v/v). Apres la sparation, et
schage des plaques, la rvlation des standards de
migration se fait par vaporisation de
dicluorofluoresceine, puis repos 5 min et puis
visualisation en UV. La zone de silice correspondant
au dpt est situe la mme hauteur que le
standard de migration plus 1 cm au dessus et au
dessous. Les bandes de silice correspondant aux
phospholipides dintrt (PE, PC, PI+PS) sont
grattes et rcupres dans des tubes en verre vis.
Transmethilation et GC (chromatographie
gazeuse)
Les acides gras sont trans mthyls en
prsence de 750l mlange tolune : mthanol (2/3,
v/v), 750 l BF3 14% suite une fermeture sous
azote, 100C dans une cuve thermostatique
pendant 90 min. La raction est arrte en plongeant
les tubes dans la glace et en ajoutant 1.5 ml de
carbonate de potassium (K
2
CO
3
) 10% afin de
neutraliser le milieu. Les esters mthyliques dacides
gras ainsi obtenus sont extraits par 2ml isooctane
pestipur ensuite fermeture sous azote. On applique
une centrifugation (5min, 1800torr/min) et on
obtient une sparation tri phasique. La phase
organique suprieure et rcupre dans des tubes en
verre puis vaporation sec sous jet dazote.
Les extraits sont repris dans un volume
disooctane puis analyss par chromatographie
gazeuse (GC) (Systme de chromatographie en
phase gazeuse couple la spectromtrie de masse
(GC Hewlett Packard Sries 6890, colonne capillaire
silice HP-5MS 30m x 0.25mm, gaz vecteur hlium
0.5 bar) [20, 21].
La quantification (en nanomoles) des acides
gras composants les queues hydrophobes de
phospholipides est faite laide des standards
internes (type 17 :0 PC et respectivement 17 :0 PE)
qui sont ajouts dans les chantillons au moment de
lextraction lipidique.
3. RESULTATS
Dans le cas de pathologies articulaires, on a
des changement locales qui favorisent les ruptures
par actions enzymatiques au niveau des structures
lipidiques du fluide synovial donnant ainsi
loxydation de phospholipides (PLs) qui entrane
ainsi des modifications de la composition de fluide
synovial en fonction de diffrents types de PLs. Ces
modifications ont t tudies par une analyse
lipidomique quantitative.
De lautre cot, loxydation des lipides
gnre des marqueurs doxydation lipidique avec
des modifications de la structure (notamment due
la perte des proprits hydrophiles-hidrophobes des
PLs qui assurent la structure discontinue du fluide
synovial sain). La prsence de ces marqueurs
doxydations lie une modification de la structure
a t tudie par une analyse lipidomique
quantitative.
81
Les analyses sur les chantillons de fluide
synovial pathologique nous ont permis dtablir par
chromatographie gazeuse (GC) la composition
lipidique du fluide synovial (en fonction du type de
PLs ainsi que pour la composition en acides gras,
saturs et non saturs, pour chaque PL identifi)
(Tableau 1).
Par chromatographie liquide haute
performance (HPLC) de mettre en vidence la
prsence dans le fluide synovial de deux mdiateurs
lipidiques dinflammation (Figure 3), associs aux
diffrentes pathologiques articulaires (ces
mdiateurs produisent la destruction enzymatique
des assemblages molculaires du fluide synovial).
Figure 3. Mdiateurs lipidiques dinflammation (HPLC, pics caractristiques dadsorption)
Pour les chantillons de fluides synoviaux
analyss, trois classes majeures des PLs ont t
identifies : PC, PE, PI+PS. Les quantits moyennes
concernant la composition en acides gras des PLs
analyss sont prsentes dans le Tableau 2. En
utilisant les rsultats obtenus par GC, on peut
constater que pour les acides gras non saturs lacide
olique (C18 :1) est le plus abondant acide gras.
Pour le comportement tribologique
articulaire, la consquence de cette variabilit de la
composition chimique des fluides synoviaux tudis
peux tre identifi dune cot, en fonction de type de
phospholipides (ayant des petite ou grosses ttes,
chargs ou neutres) quand on peut avoir une
modification de laccrochage au niveau des
assemblages lipidiques et de lautre cot, en fonction
de type dacides gras constituant les queues
phospholipidiques. Dans ce cas on distingue une
faible (cas dacides gras saturs, en phase solide) ou
une forte (pour les acides gras non saturs, en phase
fluide) mobilit des lipides lintrieur de ces
structures et donc une modification de leurs
rhologie. Ces modifications locales peuvent
entraner des modifications du comportement
tribologique pour le cas des articulations synoviales
pathologiques.
Lanalyse des rsultats obtenus ne permet
didentifier les variations pathologiques dans le cas
des chantillons tudis :
pour le phospholipide de type PE
On remarque une quantit plus leve dans le
cas ALI , on a donc + des charges au niveau des
assemblages lipidiques du fluide synovial ce qui
entrane une modification de laccrochage ;
De plus, un % suprieur en acides gras non
saturs (pour ALI et AG ) indique une plus
grande mobilit des structures lipidiques et donc une
modification de la rhologie
pour le PC
On remarque quil est le phospholipide le
plus abondant dans la composition de fluide
synovial ; sa structure (ttes petits et neutres) ne
dtermine pas des modifications significatives de
laccrochage
Il y a un quilibre entre le % en acides gras
saturs et celui en acides gras saturs donc, on nas
pas non plus des changements nets des phases des
assemblages lipidique et non plus de leurs rhologie
pour les PI+PS
On remarque une quantit plus leve dans le
cas ALI , on a donc + des ttes + volumineuses
au niveau des assemblages lipidiques du fluide
synovial ce qui entrane une modification de
laccrochage ;
Pour toutes les chantillons analyses on a
obtenu un % suprieur en acides gras saturs, ce qui
caractrise une faible mobilit des assemblages
lipidiques et donc pas de changements nets de leurs
phases et leurs rhologie.
82
Tableau 1. Variation de la composition lipidique des fluides synoviaux pathologiques
a. en fonction de type de phospholipide
Phospholipides I AL ALI AG
PE (nanomoles)
PC (nanomoles)
PI+PS (nanomoles)
37
434
234
53
546
347
84
588
356
78
472
272
b. en fonction des acides gras saturs ou non saturs composant les phospholipides
Phospholipides Acides gras I AL ALI AG
PE
PE
PC
PC
PI+PS
PI+PS
Saturs (w%)
Non saturs (w%)
Saturs (w%)
Non saturs (w%)
Saturs (w%)
Non saturs (w%)
42
53
52
46
62
37
37
59
50
49
59
40
27
68
48
50
61
38
23
70
45
53
51
49
Tableau 2. Composition en acides gras de principales classes de PLs prsentes
dans les fluides synoviaux pathologiques analyss
Acides gras % PC total % PE total %PI+PS total
12:0 0,00 0,00 0,00
14:0 0,17 0,40 0,10
16:0 DMA 3,95 0,66 0,96
16:0 16,33 34,22 25,94
16:1 1,10 1,20 0,70
17:0 0,00 0,00 0,00
18:0 DMA 3,83 0,21 0,55
18:1 DMA 1,23 0,08 0,15
18:0 14,55 16,30 15,63
18:1 n-9 11,82 14,29 6,03
18:1 n-7 2,46 0,28 0,17
18:2 DMA 0,00 0,00 0,00
18:2 n-6 8,05 15,17 4,05
20:0 0,09 0,05 2,05
18:3 n-6 0,16 0,03 0,01
18:3 n-3 0,14 0,05 0,04
20:1 n-9 0,21 0,15 0,07
20:2n-6 1,55 0,35 1,17
20:3n-9 0,00 0,01 0,00
20:3 n-6 1,10 3,45 0,40
22:0 0,17 0,05 6,18
20:4 n-6 18,49 9,11 4,48
22:1 0,06 0,02 0,19
20:5 n-3 0,52 0,46 0,08
24:0 0,76 0,13 7,75
22:2n-6 1,83 0,06 1,65
22:4n-6 3,73 0,43 0,43
24:1 n-9 0,08 0,04 16,66
22:5n-6 0,33 0,11 0,01
22:5 n-3 1,93 0,48 0,75
22:6 n-3 5,38 2,23 3,82
TOTAL acides gras
saturs (%)
32,06 51,15 57,65
TOTAL acides gras
non saturs (%)
58,92 47,90 40,69
83
Le fluide synovial fait lobjet de nombreuses
tudes cherchant identifier les composants qui
assurent la lubrification en rgime limite (boundary
lubrication) [5]. Il est montr dans la littrature que
les PLs sont impliques dans le mcanisme de la
lubrification au niveau des articulations [2,12], des
poumons [22], du pricarde [5].
Pour le comportement tribologique il est donc
fort important danalyser la lubrification en fonction
des diffrents paramtres comme: la longueur des
chanes des acides gras, leurs orientation, les tailles
des ttes hydrophiles des PLs, laccrochage au
niveau des surfaces au niveau du contact etc.
Cette tude se concentre sur la composition
lipidique en PLs, par rapport leurs chanes dacides
gras (longueur et saturation) en utilisant des
chantillons de fluide synovial pathologiques et les
ttes hydrophiles (PC, PE, PI et PS).
Les principales classes de PLs identifies
taient : PC (32 %) le composant majoritaire, et PE
(20%) et Pi+PS (20%) prsentes en quantits
signifiantes. Sarma et al ainsi que Wthier et al ont
mentionne des rsultats similaires [13, 23].
Les objectifs taient dtudier les ventuelles
modifications en composition en fonction de ltat
pathologique du fluide synovial ainsi que de mettre
en vidence la prsence des marqueurs de stress
oxydant dans la composition de fluide synovial
pathologique (diffrentes cas cliniques ont t
tudis).
Les rsultats de cet tude nous ont montr les
variations de la composition lipidique de diffrentes
fluides synoviaux pathologiques en fonction du type
de phospholipide contenu ainsi quon fonction des
acides gras formant leurs queues phospholipidiques
(par analyse en GC et HPCL).
Cela nous permet maintenant de reproduire
ex-vivo les variations de ces compositions et donc
dtudier par la suite linfluence de ces variations sur
laccrochage et sur la rhologie dans le cas des
fluides synoviaux pathologiques, afin de dterminer
les modifications du fonctionnement tribologique
articulaire pathologique.
Par la suite ces rsultats nous permettront de
reproduire ex-vivo et de manire raliste, les
pathologies articulaires afin de comprendre au cours
des essais tribologiques utilisant ce modle les
diffrents comportements et lvolution tribologique
des assemblages lipidiques en fonction des
pathologies.
4. CONCLUSIONS ET PERSPECTIVES
Lanalyse lipidomique nous a permit
didentifier les variations de la composition et les
modifications de la structure pour le cas des fluides
sinoviaux pathologiques ; on connat donc
maintenant les paramtres de disfonctionnements
pathologiques qui caractrise le modle ex des
pathologies articulaires. Lexploitation de ce modle
au cours des tudes ex vivo nous permettra de
comprendre lenchanement causes (pathologies)
consquences (symptomes) dans le but de proposer
des solutions doptimisation des traitements pour les
pathologies articulaires (mdicaments et/ou
implants)
En conclusion, PC (32%), PE (20%), PI+PS
(20%) sont les trois principales classes de PLs
prsentes dans le fluide synovial pathologique. Avec
le PC en tant que constituant prdominant. De plus,
cette tude a montr la prsence dun mlange des
acides gras au niveau des queues lipidiques des PLs
analyses, avec un % suprieur pour les acides gras
non saturs (59%) par rapport aux acides gras
saturs (32%).
En pratique, pour traiter efficacement
larthrose aprs avoir identifier chaque
disfonctionnement biologiques, il faut en connatre
les effets mcaniques et physicochimiques afin de
pouvoir dvelopper des mdicaments capables:
dagir principalement sur llment du triplet
tribologique qui est la cause de la pathologie,
dtre efficaces dans des conditions
physicochimiques variables, imposes par
lvolution de la maladie.
Des tudes supplmentaires serraient
ncessaires pour avoir une caractrisation dtaille
de linteraction des phospholipides au niveau de la
surface de cartilage, permettant ainsi une meilleure
comprhension sur la lubrification et les mcanismes
de frottement. Cela pourrait inclure la quantification
sparment des autres espces molculaires de type
PL, la quantification de la sphingomiline et du
cholestrol, sachant que leurs prsence influence la
mobilit des assemblages molculaire contenant les
lipides, lorientation des lipides dans les liposomes
pour avoir une information sur lorientation au
niveau des surfaces daccrochage, ltude des
interactions lipides protines et la modlisation
molculaire pour estimer lassemblage ainsi que
lorientation des PLs au niveau des bicouches.
5. REMERCIEMENTS
Les auteurs veulent remercier au collectif du
bloc opratoire Orthopdie adultes de lHpital
Edouard Herriot (Lyon), pour nous avoir fourni les
chantillons de fluides synoviaux pathologiques tant
ncessaires pour mener fin cette tude.
REFERENCES
1. Hills, B.A., 1998, Enzymatic Degradation of the
Load Bearing Boundary Lubricant in Joint, Br J
Rheumatol, 37, pp. 137-142.
2. Schwartz, I.M. and Hills B.A., 1998, Surface-
Active Phospholipide as the Lubricating Component
of Lubricin, Br J Rheumatol, 37, pp. 21-26.
84
3. Furey, M. J., and Burkhardt, B. M., 1997,
Biotribology: Friction, Wear and Lubrication of
Synovial Joints, Lubr. Sci., 9, pp. 255271.
4. Furey, M. J., 1997, Exploring Possible
Connections Between Tribology and Osteoarthritis,
Lubr. Sci., 9,pp. 273281.
5. Hills, B.A., 1984, Surfactants Identified in
Synovial Fluid and Their Ability to Act as Boundary
Lubricants, Ann Rheum Dis, 43, pp. 641-648.
6. Hills, B.A., 1990, Oligolamellar Nature of the
Articular Surface, J Rheumatol, 17, pp. 349-356.
7. Hills, B.A., 1989, Oligolamellar Lubrication of
Joints by Surface Active phospholipid, J
Rheumatol, 16, pp. 82-91.
8. Schmid, R. et al, 2001, "Alteration of Fatty Acid
Profiles in Different Pulmonary Surfactant
Phospholipids in Acute Respiratory Distress
Syndrome and Severe Pneumonia, Am. J. Crit. Care
Med., 163, pp. 95-100.
9. Jay, G.D, Britt, D.E., Cha, D.J., 2000,
Lubricin is a Product of Megakaryocyte
Stimulating Factor Gene Expression by Human
Synovial Fibroblasts, J Rheumatol, 27, pp.594-600.
10. Momberger, T.S., Levick, J.R., and Mason,
R.M., 2005, Hyaluronan Secretion by Synoviocytes
is Mechanosensitive, Matrix Biol, 24, pp.510-519.
11. Schumacher, B.L., Block, J.A., Schmid, T.M.,
1994, A Novel Proteoglycan Synthesized and
Secreted by Condrocytes of the Superficial Zone of
Articular Cartilage, Arch Biochem Biophys, 311,
pp. 144-152.
12. Trunfio-Sfarghiu, A.-M., Berthier, Y.,
Meurisse, M.-H., Rieu, J.-P., 2007, Multiscale
Analysis of the Tribological Role of the Molecular
Assemblies of Synovial Fluid. Case of a Healthy
Joint and Implants, Tribology International, 40, pp.
1500-1515.
13. Sarma, A. V., and Powell G. L., 2001,
Phospholipid Composition of Articular Cartilage
Boundary Lubricant, Journal of Orthopaedic
Research, 19, 671-676.
14. Blewis, M. E. et al, 2007, Model of Synovial
Fluid Lubricant Composition in Normal and Injured
Joints, European Cells and Materials, 13, pp. 36-39
15. Gale, L.R., Chen, Y., Hills, B.A., Crawford
R., 2007, Boundary Lubrication of Joints, Acta
ortopedica, 78, pp. 309-314.
16. Bordet J.-C., Guichardant M., Lagarde M.,
1990, Modulation of Prostanoid Formation by
Various Polyunsaturated Fatty Acids During
Platelet-Endothelial Cell Interactions,
Prostaglandins Leukot Essent Fatty Acids, 39, 197-
202.
17. Corneci M.C. et al., 2007, "Optimisations des
surfaces frottantes des implants articulaires pour
favoriser la lubrification par les assemblages
lipidiques", 34TH LEEDS-LYON SYMPOSIUM
ON TRIBOLOGY SUMMER SCHOOL, Lyon.
18. http://www.cyberlipid.org/
19. Wolff, J.P., 1968, "Analyse des lipides et
sparation des acides gras par CCM"
20. Benzaria, A., 2006, Etude biochimique et
nutritionnelle de l'effet immunomodulateur des
huiles de poisson, d'olive et d'argan. Effets
compars de leurs acides gras, Thse : Institut
National des Sciences Appliques de Lyon,
[26/02/2007], p. 105-106.
21. Goerke J, 1998, Pulmonary Surfactant:
Functions and Molecular Composition, Biochim.
Biophys. Acta, 1408, pp. 7989.
22. Soares, A.F., 2005, Effets du stress oxydant sur
le fonctionnement des adipocytes : adiponectine et
prostaglandines [En ligne]. Thse : Institut National
des Sciences Appliques de Lyon, 2005
[04/05/2006], 130 p. Disponible sur :
http://docinsa.insa-lyon.fr/these/pont.php?id=soares
23. Wthier, R.E., 1968, Lipids of Mineralizing
Epiphyseal Tissues in the Bovine Fetus, Journal of
lipid research, 9, pp. 68-78.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 85-88
Ionut Cristian ROMANU
email: ionutromanucristian@yahoo.com
Emanuel DIACONESCU
email: emdi@fim.usv.ro
Department of Applied Mechanics,
University Stefan cel Mare of Suceava,
ROMANIA
BIOARTICULAR FRICTION
The present paper illustrates experimental investigations of
bioarticular friction. The first set of experiments was conducted on
a pig synovial joint and the second one investigates the friction
between a spherical cap made out of cartilage and an elastic half-
space. For the experimental investigations, a device was conceived
and built that ensures rolling and sliding movements of the joint.
Keywords: bioarticular friction, friction coefficient, synovial joints
1. INTRODUCTION
The human body contains 143 joints that
connect skeletal bones [1]. Most of these joints are
synovial and represent the object of present study.
Human synovial joints are subjected to various and
large forces under static and dynamic loading, while
executing sliding and rolling movements [2].
A joint represents the connection between
two or more bones, via a fibers and ligaments.
Viewed from the mechanical engineers perspective,
joints can be treated as cinematic couplings. A
healthy joint must ensure a series of functions such
as [3]:
allowing bone movements in particular
directions;
ensuring low friction between contacting
surfaces;
receiving and transmitting forces;
shock and vibration damping.
2. JOINT TYPES
The first to classify the joints into categories
was Bichat [1]. Using physiological characteristics
as a criterion, Bichat suggested that joints are either
mobile joints (later named diarthrosis by Galien ) or
fixed joints (also named synarthrosis). Another type
of joint can be distinguished, having less mobility,
also called amphiarthrosis.
3. EXPERIMENTAL SET-UP
Experimental investigations were conducted
on a joint from an approximately 250 days old pig.
For preservation, the joint was kept in a container
filled with 0.98% saline solution, at constant
temperature and in darkness.
Figure 1. Types of joints: a) Synarthosis, [2];
b) Diarthrosis, [2]; c) Amphiarthrosis, [4]
Mechanically, the joint was clamped on an
adjustable arm at one end and to a stiff enough
elastic lamella at the other. The elastic lamella is
rigidly bound to the mobile core of an electro-
dynamic actuator. Once the actuator is turned on,
the joint is subjected to both rolling and sliding
movements.
Besides the mechanical part, ensuring joint
movements, as shown in Figure 2, the experimental
set-up consists of the following auxiliary
components: signal generator, audio amplifier, strain
indicator, oscilloscope and PC.
The devices used in the experimental
installation are connected one to another as
illustrated in Figure 3.
86
Figure 2. Experimental set-up [6]
Figure 3. The connections between devices
The first step of presented experimental
investigations was calibration of the measuring
devices. To this end, the strain indicator was first
set to indicate 0 when subjected no load, and a
sinusoidal signal with a frequency of 3 Hz and an
amplitude of 0.8V was programmed at the signal
generators output. In order to quantitatively assess
forces transmitted in joints, the elastic lamella was
loaded using dead weights of known mass. The
values indicated by the strain indicator are presented
in Table 1.
After calibrating the measuring apparatus, the
joint was fixed into position on the device as
described before, and measurements were taken
under different conditions.
Table 1. Loading device calibration values
Weight
[g]
Strain
indicator
value
[mV/V]
Equivalent
load [N]
5 0.002 0.05
10 0.004 0.1
20 0.007 0.2
25 0.008 0.25
30 0.01 0.3
50 0.016 0.5
100 0.031 1
150 0.046 1.5
200 0.061 2
Experimental measurements of friction were
made for several different frequencies, in the form of
oscilloscope graphical charts, as the one illustrated
in Figure 5.
In order to calculate joint friction, the 2Hz
chart was considered. This frequency corresponds
to a 1.80m tall man having an average step width of
0.8m / step, at a frequency of 2 steps / s. This means
he would walk at a speed of 1.6m / s or 5.7 km / h,
which represents the average speed at which a
normal human is moving most of the time.
87
Figure 4. Joint clamped on the experimental
apparatus [6]
Figure 5. Friction curve at 2Hz
The graph in Figure 5 indicates that peak to
peak signal amplitude is 410mV, and 205mV for a
half alternation respectively. By comparing the
charts with values obtained when the device was
statically loaded for calibration, it is shown that for a
1N load, an output voltage of 220 mV is generated.
In this manner, it is possible to graphically conclude
that maximum joint friction occurs when the signal
reaches a maximum. For such maximum the force
can be evaluated as shown in eq. (1):
f
1 205
F 0.931N
220

. (1)
Thus, for a 2 Hz frequency, the friction force
evaluated at 0.931N. This measured force is the
result of both cartilage friction and loss due to
friction between articular tissue linings.
To better assess friction force between bone
ends in a joint, a second experimental rig described
in [6, 7] was employed to study the contact between
a bone end and a flat glass surface. The
investigations are based on contact mechanics
theory, according to which the contact between two
spherical punches (bone ends in this case) can be
replaced by the contact of an equivalent punch
pressed against a half-space (represented by the
glass plate).
Figure 6. Second experimental set-up [7]
The second experimental set-up consists of a
mechanical part that ensures loading and movement
of the bone pressed against a glass plate, strain
indicator, oscilloscope and PC.
As before, the test rig was first calibrated,
obtaining for the elastic lamella used in friction
measurements a calibrating curve as the one shown
in Figure 7.
0 0.5 1 1.5
0
0.2
0.4
F
u
Figure 7. Elastic lamella calibration curve
In the experimental investigations, the
following steps were covered:
Cleaning the glass disc;
Connecting the experimental device to the
power supply and strain indicator;
Interfacing oscilloscope and strain indicator;
Calibrating strain indicator;
Equipment functionality was checked before
fixing bone end;
Fixing bone end on the support by screw;
Lowering the glass disc until contact with
bone end is established.
Operating the test rig, thus creating bone end
movements, and recording friction charts as the one
depicted in Figure 8.
The chart in Figure 5 indicates a peak to peak
signal amplitude of 0.8V, value obtained during
calibrations at a 20 g weight, which corresponds to a
0.2N applied force.
In order to validate the experiments, friction
coefficient was calculated and compared against
literature values.
88
Figure 8. Friction curve at 15 N loading
Thus, for a friction force of
f
F 0.02 N at a
N 15 load, the resulting friction coefficient is:
f
F
0.013
N
. (2)
This value was found to be close to friction
coefficient values from literature.
Another set of tests involved placing water
between the bone end and glass plate, thus better
modeling real life lubrication conditions. A typical
resulting friction curve is shown in Figure 9.
Figure 9. Friction curve (with water lubrication)
According to the chart, signal amplitude is
0.6V, which corresponds to a static load of 0.15N,
leading to a friction coefficient value as follows:
f
F
0.01
N
. (3)
From these results it can be concluded that, in
the presence of a lubricating layer (in this case
water), friction decreases up to 70% from its initial
value. Water was chosen as lubricant because
synovial fluid is 90% water.
4. CONCLUSIONS
The work reported here can be summarized
by a few conclusions listed below.
The first experiment aimed to determine joint
friction without separating joint elements. Several
measurements were taken at different frequencies
and friction force was calculated at 2 Hz, frequency
corresponding to a 1.8 m tall man walking at 5.7km
per hour. For these conditions, a friction force of
0.931N was determined in the joint.
At first glance, such friction value may seem
high, but this value includes the friction between
cartilage covered bone ends and that between tissues
lining the joint.
Graphical results show an increase in friction
with frequency, which can be attributed to intra-
articular fluid viscosity.
For a better assessment of friction, a second
test rig was employed, in which cartilage covered
bone ends moves against a flat glass surface. From
this second experiment, friction forces were
evaluated at 0.2 N for dry contact and at 0.15N when
a liquid (water) was used as lubricant respectively.
On the second rig, measurements were taken
at the same frequency (given by the motor drive of
the device), but contact load varied.
A linear dependence of friction to load was
observed.
In order to validate obtained results, friction
coefficients were calculated and found to be in good
agreement with literature values.
REFERENCES
1. Bichat, X., 1829, Anatomie descriptive, JS
Chaud, Paris, tome II, pp 41-44
2. Merkher, Y., Sivan, S., Etsion, I. , Maroudas,
A., Halperin, G., Yosef, A., 2006, A Rational
Friction Test Using a Human Cartilage on Cartilage
Arrangement, Parma.
3. Diaconescu, E., Glovnea, M., Bejinariu, B.,
2008, Experimental Evidence Upon Contact
Behavior of Cartilage Covered Bone Ends,
Proceedings of VAREHD 14, Suceava;
4. Furey, M.J., 1996, Friction Wear and
Lubrication of Natural Synovial Joints, Proc. Of
12th int. Colloquium on Tribology, 1421-1430,
Esslingen.
5. Diaconescu, E., Glovnea, M., Frunza, G.,
2007, Metode inovative de bio-ortopedie pentru
reconstrucia articular, (in Romanian), contract
CEEX BIOART Nr. 70 / 2006, Suceava.
6. Romanu, I., 2009, Bioarticular friction, (in
Romanian), Graduation Paper, University of
Suceava.
7. Brndua, B., 2008, Experimental modeling of
bioarticular contacts, (in Romanian), Graduation
Paper, University of Suceava.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 89-105
A.-M. TRUNFIO-SFARGHIU
1
e-mail: ana-maria.sfarghiu@insa-lyon.fr
M.C. CORNECI
1,2
Y. BERTHIER
1
M.-H. MEURISSE
1
J.-P. RIEU
3
1
Laboratoire de Mcanique des Contacts et des
Structures, INSA-Lyon, CNRS UMR5259,
F69621 Villeurbanne Cedex, FRANCE
2
Universit Technique Gh. Asachi, Facult de
Mcanique, 700050, Iasi, ROUMANIE
3
Laboratoire de Physique de la Matire
Condense et Nanostructures, Universit Claude
Bernard Lyon 1, CNRS UMR5586, F69622
Villeurbanne Cedex, FRANCE
MECHANICAL AND PHYSICOCHEMICAL
ANALYSIS OF THE TRIBOLOGICAL
OPERATION OF JOINT REPLACEMENTS
The aim of this work is to identify the coupled role of the biological
components of synovial fluid in the remarkable tribological
operation of a healthy natural joint, as well as in the friction of steel
and polythene implants. It uses a realistic ex-vivo model capable of
reproducing the mechanical and physicochemical characteristics of
the entire tribological triplet of the joint, whether healthy or
implanted. It particularly focuses on the lipidic bilayers and vesicle
structures associated with synovial fluid. The analysis of the
friction measurements and fluorescence microscopy images confirm
the role of lipidic bilayers in maintaining a very low friction
coefficient. In addition, we observe that the substitute cartilage
favours the formation and maintenance of these bilayers, which is
not the case of implant materials.
Keywords: synovial joint, biolubrication, molecular assemblies,
lipidic bilayers, articular cartilage
1. INTRODUCTION
Over the years the growing number of osteo-
articular diseases has led to the development of joint
implants whose lifetimes depend on their
tribological performances. In spite of sustained
efforts to develop new biomaterials, the in-vivo
lifetime of implants has often proven to be most
deceptive when set against extrapolations performed
on the basis of ex-vivo simulations. This
discordance can be imputed mainly to ex-vivo
tribological test conditions that are insufficiently
realistic in comparison with the mechanical and
physicochemical conditions of biological
environments.
This is why interest has grown over the last
few decades in attempting to identify and
characterise the biomolecular interfaces formed
during the tribological operation of healthy and
implanted joints. Thus four biological components
of synovial fluid have been identified as being
decisive in the tribological performances of joints
and certain works have attributed separate roles to
them:
albumin protects against wearing of artificial
cartilage [1] and the metal surfaces of joint implants
[2],
hyaluronic acid tends to increase the viscosity
of healthy synovial fluid [3] at low shear rate. But at
the high shear rates encountered in synovial joints,
the effect of hyaluronic acid on apparent viscosity is
not significant [4].
lubricin and other polyelectrolytes attach to
the surface of healthy cartilage and modify
tribological conditions under boundary regime [5,6].
phospholipids multilayers play also an
important role in the boundary lubrication regime
[7,8].
However, these works have neglected the role
of the multiple interactions between the components
of the synovial fluid and the bodies in contact during
the tribological operation of the joint. Recent
research has suggested the essential tribological role
of such interactions.
Under tribological stress the albumins create
reticulations with the hyaluronic acid that modify the
rheology of the synovial fluid [9].
The association of hyaluronic acid with lipids
can form pocket and tube-like structures filled with
hyaluronic acid and surrounded by lipidic
multilayers [10]. Furthermore, it has been shown
that such structures reduce the rheothinning effect of
aqueous solutions of hyaluronic acid [11].
The vitonectin part of lubricin can fix
lipidic bilayers on articular cartilage while its
hemopexin part protects them against oxidation
[12-14].
The healthy joint is an ultra-high
performance tribological system and an essential
reference for understanding the tribological
operation of synovial fluid and thus aiding research
aimed at improving the treatment of joint diseases
and optimising implant design. Consequently, this
work is based on a realistic ex-vivo model to
understand the coupled role of the molecular
90
components of synovial fluid involved in the
remarkable tribological performances of natural
joints. It also examines the potential capacities of
these components to improve the tribological
performances of joint implants.
The ex-vivo model proposed here, and
presented in section 2, is designed to reproduce the
mechanical and physicochemical characteristics of
the whole tribological triplet, either healthy or
implanted, composed of first bodies in contact
(articular cartilage and implant materials), the third
body (the synovial fluid with its real biomolecular
structure) which separates the first bodies, and
obviously the mechanism (muscle and ligament
system) which imposes local loadings. This article
focuses on the boundary regime under which the
effects of hydrodynamic load carrying capacity are
completely negligible, and which exacerbates the
role played by the interfaces.
The tribological analysis proposed in section
3 is based on optical microscopy observation of
lipidic interfaces performed under white and
fluorescent light, in situ through a transparent first
body, and after opening the contact. The changes in
the lipidic structures are correlated with the changes
of the friction coefficient.
2. EX VIVO MODEL TRIBOLOGICAL
TRIPLET
This section presents a model of an ex-vivo
tribological triplet of a natural and implanted joint,
with realistic mechanical and physicochemical
parameters. We successively describe the elements
making up the tribological triplet: the "first bodies"
(substitute cartilage and implant material), the "third
body" (substitute synovial fluid), and the
"mechanism" (experimental device providing
contact pressures and kinetics). The entire
procedure used to define this experimental model is
presented in [15].
2.1 First bodies
2.1.1 Substitute cartilage
Taking samples of cartilage affects the
complexity of its structure and in particular destroys
the collagen membrane on its surface. What is
more, it loses its living properties when subjected to
long-term tribological tests, thereby leading us to
seek non-living materials capable of forming a
substitute joint cartilage that corresponds as much as
possible to the mechanical and physicochemical
properties of real cartilage. Consequently, we turned
to highly hydrophilic polymeric materials like
hydrogels, since cartilage contains 80% water in
volume. These materials are used to repair cartilage
injuries: polyalcohol vinyl (PVA) and hydroxyethyl
methacrylate (HEMA) [16,17,18]. Of the latter we
opted for the hydrogel HEMA used for corneal
lenses, since its structure and mechanical properties
are very close to those of cartilage:
HEMA hydrogel has large methacrylate
polymeric chains reticulated by hydroxy-ethyl
groups (Figure 1a) that are hydrophilic due to their
negative HO- charges. This structure is comparable
to that of cartilage which has collagen fibres
reticulated by glucidic chains (aggrecan, see Figure
1b), that are hydrophilic due to their negative SO3-
and COO- charges..
Table 1 highlights the similitude between the
mechanical properties of HEMA hydrogel after 48
hours in physiological solution and articular
cartilage.
The first bodies used as substitute for joint
cartilage are semi-rigid blanks of HEMA lenses
(Corneal, France) whose macro-geometry has a
domed part (cf. Figure 1a.), making it possible to
localise the contact during the friction tests. The
roughness of this contact zone is very low (Ra of a
few nanometres), comparable to the one of natural
cartilage compressed and flattened to nanometer
levels under the pressures in human hip and knee
joints [19].
a. b.
Figure 1. Schematic representation of hydrogel and cartilage structure: a) hydrogel HEMA (at macroscopic and
molecular scales), b) cartilage (at macroscopic and molecular scales) from [11]
91
Table 1. Mechanical properties of hydrogel HEMA and cartilage listed from literature values
Hydrogel HEMA [21] Cartilage [22]
Equilibrium compressive modulus (MPa) 0.2 0.9 0.5 - 1
Permeability (m
4
/N.s)
measured for a gradient of hydrostatic pressure of
21 MPa through 1 mm
~ 10
-16
10
-16
- 10
-15
Water content (% mass)
~ 25 %
External layer ~ 20 %
Internal layer ~70 %
Since this hydrogel reproduces only partially
the structure and some mechanical properties of the
real cartilage, further on it will be called artificial
cartilage, following the term of Murakami [20].
2.1.2 Joint implants
Two types of material used in joint implants
were considered: polyethylene (UHMWPE) and
stainless steel (316L). Cylindrical test pieces 15 mm
in diameter and 3 mm thick had an RMS roughness
of 0.1 m for UHMWPE and 0.05 m for stainless
steel, representative of the roughness of the rubbing
surfaces of joint implants.
2.2 Model of synovial third body
The model of the synovial fluid used in this
study is based on histological and biophysical
observations showing that synovial fluid forms
vesicles (or "pockets") [23] several micrometers in
diameter filled with a gel composed of hyaluronic
acid and albumins (Figure 2a). According to [10],
these "vesicles" are coated with lipidic multilayers
formed by the stacking of lipidic bilayers and layers
of physiological solution that can also be found at
the interface between the synovial gel and the
cartilage [8] (Figure 2b).
This type of structure has been reproduced
ex-vivo by using recent technology derived from
nanostructural physics and described below.
2.2.1 Formation of synovial gel vesicles
Synovial gel was synthesised from DPPC
lipids (1,2-Dipalmitoyl-sn-Glycero-3- Phopho-
choline, 850355CP Avanti Polar Lipids), hyaluronic
acid (H7630 Sigma-Aldrich) and serous albumin
(A1653 Sigma-Aldrich), acquired in powder form,
then placed in physiological solution, in order to
obtain concentrations equivalent to those of synovial
fluid.
Accumulations of lipids due to their very low
solubility in aqueous solutions were eliminated by
using a technique specific to the fabrication of
liposomes [24]. They first consist in generating two
solutions with identical volumes:
a solution of 3 g/l lipids in a solvent
composed of 90% chloroform and 10%
ethanol in volume,
a solution of 3 g/l hyaluronic acid and 20 g/l
serous albumin in physiological solution.
The gel vesicles are then formed by following
the procedure schematized in Figure 3, whose
successive steps are as follows:
evaporation of the solvent in nitrogen (Figure
3a) and centrifugation of the first solution
causes the lipids to spread over the internal
wall of a glass test tube.
the hyaluronic acid and serous albumin
solution is then added to the test tube and the
resulting mixture is subjected to ultrasound
for 2 minutes to trigger the formation of the
vesicles, before leaving them to incubate for
48 hours at 45C so that the vesicles
incorporate the synovial gel (Figure 3b).
a) b) c)
Figure 2. Schematic view of the synovial fluid : a) discontinuous structure made up of lipidic pockets filled of
synovial gel (hyaluronic acid + albumine), b) lipidic multilayered structures at the interface between synovial gel
and cartilage, c) detail of the lipidic multilayered structures fixed on articular cartilage by lubricin
92
Figure 3. Procedure of formation of vesicles of synovial gel: evaporation of solvent to cover lipids on the
surface of the test tube, suspension in synovial gel (hyaluronic acid + albumine), sonication to form small
unilamellar vesicles and incubation to form lipid pockets filled with synovial gel (see methods for more details)
2.2.2 Adhesion of lipidic bilayers on the surface of
the first bodies
To stimulate ex-vivo the multilayer properties
of the interface between the first and third bodies in
a healthy joint as well as possible (Figure 2b), we
initiated the adhesion of lipidic bilayers on the first
bodies by using a lipidic vesicle fusion method
[25,26].
This first entails forming lipidic bilayers
several hundred nanometres in diameter, by
subjecting an aqueous suspension of 2 g/l lipids to
ultrasound (at 50W power) for 5 minutes. This
suspension, diluted ten times, is then used to
produce the lipidic deposit.
As shown in the diagram in Figure 4, the
deposit technique consists in:
leaving the surfaces of the first bodies to
incubate for 5 minutes in a diluted suspension
of small lipidic vesicles, to which 2mmol/l of
Ca++ ions were added to stimulate the
vesicles to adhere and burst on the surfaces of
the first bodies (Figure 4a),
eliminating the lipidic surplus by rinsing
(Figure 4b).
This technique permits obtaining model
lipidic bilayers containing Ca++ ions. The work
done by Hills [18, 22] has shown that these ions are
also present in in-vivo lipidic bilayers where they
stiffen the bilayer by forming ionic links with the
negative parts (phosphate group) of the lipidic heads
(Figure. 4c).
Figure 4. Formation of lipidic bilayers by the vesicle fusion method: a) incubation, adsorption and fusion of
vesicles, b) elimination of the lipidic surplus by rinsing, c) internal structure of lipidic bilayers (from [27])
a. b.
Figure 5. Chemical-molecular structure of lipids: a) 1,2-Dipalmitoyl-sn-Glycero-3-Phosphocholine (DPPC), b)
1-Palmitoyl-2-[12-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl]-sn-Glycero-3-Phosphocholine
(16:0-12:0 NBD PC)
93
2.2.3 Viewing lipidic structures
Molecular markers were added at 1% molar
concentration in DPPC lipidic powder in order to
view the initial structure of the third body and its
evolution during the friction tests. These are
NBDPC lipids (Avanti Polar Lipids), whose ends are
fluorescent under blue light (Figure 5).
In order to focus the microscope on a
transparent interface (lipidic bilayer, glass surface,
free surface of a solution of synovial vesicles),
focusing was done in white light on the projection
on this surface of the octagonal contour of the
microscopes field diaphragm. This focusing was
then kept when changing to blue light to view the
fluorescent elements of this interface [28].
Several situations can be observed under blue
light:
If the third body does not contain any
fluorescent elements under blue light, the diaphragm
does not appear in the image. This is, for example,
the case of a glass surface without a lipidic bilayer,
or of the free surface of physiological solution
(Figure 6a).
If the third body contains fluorescent
elements, the images obtained under blue light are
composed of a clear zone bordered by the octagon of
the diaphragm. This clear zone can be uniform, as
in the case of a glass surface with an intact lipidic
bilayer (Figure 6b), or an aqueous solution of small
lipidic vesicles, or reveal details as in the case of a
solution of large vesicles of synovial gel (Figure 6c).
2.3 Experimental set-up
An experimental set-up permitting the in-situ
visualisation of the contact was developed (Figure 7)
to simulate the tribological operation of the model
articular contact.
The model first bodies (HEMA, PE and steel
samples) were fixed to the bottom of a tank
containing the third body to be tested. The tank was
linked to a translation stage by a system of flexible
blades. The translation stage imposed cyclic
translation movements forwards and backwards.
A transparent opposing first body formed the
contact with the upper surface of the model first
body. Normal load was applied by gravity. An
upright microscope (Leica DMLM) linked to a
camera (Leica DC350F) for fluorescent analytical
imaging permitted viewing the contact through the
opposing first body. This observation was done in-
situ, during friction and could be done under white
and blue (fluorescence) light. Exactly the same
camera acquisition parameters were used under blue
light to compare the different quantities of
fluorescent lipids in the contact.
An eddy current position sensor measured the
deformation of the flexible blades holding the tank,
and permitted calculating the tangential force and
the friction coefficient. The sensitivity, linearity
range and sensor position were set-up so that the
uncertainty on the force measurements between -1N
and 1N was 0.0005N.
a. b. c
Figure 6. Fluorescence microscopy images of various lipidic structures containing 1% of fluorescent NBDPC
using blue light. a) glass surface without any lipidic bilayer. b) glass surface with an intact lipidic bilayer (the
projection on the surface of the octogonal contour of the microscopes field diaphragm is seen). c) solution of
synovial gel vesicles without any lipidic bilayer near a glass surface
Figure 7. Schematic view of the experimental device
94
2.4 Experimental procedure
For the tests presented in this text, normal
load was applied at 2.5N, which, by using the curve
radii of the test pieces, permitted imposing realistic
pressure conditions. The position sensor thus
allowed the indirect measurement of friction
coefficients up to 0.4, with an uncertainty in the
region of 0.0002.
Back and forth displacements were made at a
constant speed of 0.6 mm/s. On the one hand, this
low value permitted good visualization of the
contact by the optical microscope and, on the other,
a boundary type lubrication regime. The back and
forth movements were of equal duration, in the
region of ten seconds, which, given the speed,
permitted the successive visualization of the whole
length of the contact.
Several series of friction tests, each lasting 1h
(about 180 back and forth cycles), were performed.
They included 3 types of first body and 4 types of
third body.
The different combinations of first body,
aimed at simulating natural articulations and the
different types of implant, were the following:
A contact between a convex sample (curve
radius 8mm) in soft HEMA (Young modulus of
about 1.5 MPa) hydrated for 48h in physiological
solution and a flat glass (borosilicate) opposing
surface (Figure 8a). Given the load applied, the
contact, which had a diameter of 2mm, was
subjected to an average pressure of 0.3 MPa, whose
order of magnitude seems realistic in comparison to
the operating healthy knee joint in normal gait
[29,30]. In what follows, this contact is called
"artificial cartilage contact model".
A contact between a flat sample in steel 316L
and a transparent convex opposing surface (curve
radius 8mm) made of non-hydrated rigid HEMA
(Young modulus of about 1 GPa) (Figure 8b). This
circular contact of 0.8 mm in diameter was subjected
to an average pressure of 5 MPa, which is realistic
for the operating knee joint implant in normal gait
[31]. This contact is referred in what follows as
"steel joint implant contact model ".
A contact between a flat polyethylene
UHMWPE sample and a transparent convex glass
(borosilicate) opposing surface (curve radius 25.5
mm) (Figure 8c). This circular contact of 0.8 mm
diameter bore an average pressure of 5 MPa, which
is realistic for the operating knee joint implant in
normal gait [31]. This contact is referred to in what
follows as "polyethylene joint implant contact
model".
The following 4 types of third body and
interfaces were used in order to study in uncoupled
mode the tribological role of the different
constituents of synovial fluid:
physiological solution between the first
bodies not covered with lipidic bilayers, referred to
hereafter as "3rd body A", considered as the
reference 3rd body with respect to the friction
values,
a suspension of 2 g/l of small lipidic vesicles
(several hundred nm in diameter) in physiological
solution, fluorescent under blue light (cf. 2.2.2) and
referred to hereafter as "3rd body B" (fig 9a),
physiological solution between the first
bodies initially covered with lipidic bilayers (cf.
2.2.2), referred to hereafter as "3rd body C", (fig 9b),
lipidic pockets filled of synovial gel (cf 2.2.1)
between the first bodies initially covered with lipidic
bilayers, referred hereafter as "3rd body D" (fig 9c).
3. EXPERIMENTAL RESULTS
3.1 Friction measurements
Figure 10 shows the evolution of the friction
coefficient, at the start of the test and after one hour
of operation for the artificial cartilage contact model
in the presence of 3rd body A (physiological
solution). The friction coefficient was observed to be
very stable, in the region of 0.035. The rate of
variation of the friction coefficient during a cycle is
representative of the curves obtained for each of the
combinations between the first and third bodies.
However, in other configurations, the friction value
changed through time.
a. b. c.
Figure 8. Model first bodies considered in this study:
a) model of natural joint (model 1st body = hydrated HEMA, transparent 1st body = borosilicate glass),
b) polyethylene joint implant contact model (model 1st body = 316L steel, transparent 1st body = rigid HEMA),
c) polyethylene joint implant contact model (model 1st body = UHMWPE,
transparent 1st body = borosilicate glass)
95
a) 3rd body A b) 3rd body B c) 3rd body C d) 3rd body D
Figure 9. Schematic view of the third bodies and interfaces considered in the experiments
a) 3rd body A (physiological solution), b) 3rd body B (suspension of small lipidic vesicles in physiological
solution; c) 3rd body C (physiological solution between the first bodies initially covered with lipidic bilayers,
d) 3rd body D (lipidic pockets filled of synovial gel between the first bodies
initially covered with lipidic bilayers)
In the case of the artificial cartilage contact
model, the friction tests were repeated five times, the
dispersion on the friction force being 4% of the
average value. For the implant models, each of the
tests was performed twice, resulting in a maximum
variation of 7% on the friction measurement.
All the average values of the friction
coefficients obtained in this way for the 12 test
configurations corresponding to the 3 model
contacts and 4 third bodies (cf. 2.4.) at the start and
end of the tests are grouped in Figure 11.
It can be seen that the combination of lipids
with hyaluronic acid and albumins (3rd body D)
gives a higher friction coefficient (0.12), whatever
the model contact studied, than 3rd body A.
On the other hand, the comparison of 3rd
bodies B and C, in which only lipids are present,
without hyaluronic acid or albumin, with 3rd body A
gives results as a function of the model contact and
operating time. The following can be observed:
a significant decrease of the friction
coefficient for the artificial cartilage contact model
(0.035 for 3rd body A, 0.005 for 3rd body B after 1h
and 0.0015 for 3rd body C);
an initial decrease of the friction coefficient
for the steel joint implant contact model (0.07 for
3rd body A, 0.03 for 3rd bodies B and C), but a
return to the initial value after 1h of friction;
an initial increase of the friction coefficient
for the polyethelene implant contact model (0.06 for
3rd body, 0.07 for 3rd bodies B and C), followed by
a reduction to 0.05 after 1h of friction. However, this
fall is probably not directly correlated with the
presence of lipids in the third body, since it can also
be observed with 3rd body A.
3.2 Visualization
The microscopy images of the first bodies
before friction are shown in Table 2. These show
that:
the soft hydrated HEMA surfaces and rigid
non-hydrated HEMA surfaces, as well as
those in steel and glass, permitted the
physicochemical adhesion of a fluorescent
uniform lipidic bilayer (cf. 2.2.2)
however, the polyethelene surface did not
adsorb the lipidic bilayer.
The in-situ images taken during friction are
shown respectively in the first two columns of
Tables 3, 4 and 5. The images show:
a zone including the border of the contact for
the artificial cartilage contact model (Table
3).
the central zone of the contact for the implant
contact models (Tables 4 and 5), the
experimental set-up and the contact
configuration (Figure 8b and 8c) did not
permit access to the edge of the contact.
Figure 10. Typical shape of friction curves recorded with artificial cartilage in the presence of 3rd body
96
Table 2. Visualisation using different lights of the different first bodies (1-6) with or without lipids before contact. First bodies were immersed in water to keep the lipidic
bilayer integrity. 1a-6a: Visualisation in white light of the surfaces. 1b-6b: Visualisation in blue light of the surfaces without any lipidic bilayer.
1c-6c: Visualisation in blue light of the surfaces with lipidic bilayer
97
Table 3. Visualisation of artificial cartilage contact model; 1-2: in situ before and after friction; 3-4: after contact
on both first-body (hydrated HEMA and borosilicate glass). Different third-bodies were investigated: a-b:
third body B (suspension of small lipidic vesicles in physiological solution); c-d: third-body C (physiological
solution between the first bodies initially covered with lipidic bilayers); e-f: third-body D (lipidic pockets
filled of synovial gel between the first bodies initially covered with lipidic bilayers). Images were performed
with white light (a,c,e) or with blue light (b,d,f)
98
Table 4. Visualisation of steel joint implant contact; 1-2: in situ before and after friction; 3-4: after contact on
both first-body (316L steel and rigid HEMA). Different third-bodies were investigated: a-b: third body B
(suspension of small lipidic vesicles in physiological solution); c-d: third-body C (physiological solution
between the first bodies initially covered with lipidic bilayers); e-f: third-body D (lipidic pockets filled of
synovial gel between the first bodies initially covered with lipidic bilayers). Images were performed with white
light (a,c,e) or with blue light (b,d,f)
99
Table 5. Visualisation of polyethylene joint implant contact; 1-2: in situ before and after friction; 3-4: after
contact on both first-body (UHMWPE and borosilicate glass). Different third-bodies were investigated: a-b: third
body B (suspension of small lipidic vesicles in physiological solution); c-d: third-body C (physiological solution
between the first bodies initially covered with lipidic bilayers); e-f: third-body D (lipidic pockets filled of
synovial gel between the first bodies initially covered with lipidic bilayers). Images were performed with white
light (a,c,e) or with blue light (b,d,f)
100
0.03 0.03
0.06
0.07
0.035
0.05
0.07
0.035
0.07
0.015
0.05
0.07
0.005
0.07
0.0015
0.05
0.1
0.0015
0.12 0.12 0.12
0.1
0.12 0.12
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
artificial cartilage model steel UHMWPE
Friction
coefficient
3rd body A, start value 3rd body A, end value
3rd body B, start value 3rd body B, end value
3rd body C, start value 3rd body C, end value
3rd body D, start value 3rd body D, end value
Figure 11. Friction coefficients at starting and end of each test
They permit observing changes in the
distribution of fluorescence in the contact and
possibly the exterior during friction.
The last two columns of these tables concern
the images of the first bodies at the end of the tests,
after opening the contact and rinsing with distilled
water.
In section 3.3 we propose an analysis of the
tribological role of the components of the different
third bodies tested, in each of the model contacts, by
using the correlations between the values and
evolutions of the friction coefficients and the
evolution of the images in white and blue light.
3.3 Interpretation
3.3.1 Artificial cartilage contact model
3rd body B (physiological solution, small
lipidic vesicles)
The presence of small lipidic vesicles
(approximately 200 nm in diameter) solution within
3rd body B generated quite significant fluorescence
(Table 3, image 1b, part on right) provided by
vesicles confined in the contact. This fluorescence
was a little less significant than that recorded outside
the contact (Table 3, image 1b, part on left) since the
lipidic vesicles here were not confined in the contact
and thus the thickness of the volume visualized was
greater. After friction of 1h, the contact zone
appeared much less fluorescent (Table 3, image 2b).
The reduction of fluorescence in the volume
of the 3rd body occurred along with the appearance
of fluorescence on the surface of first bodies:
uniform fluorescence on the HEMA surface (Table
3, image 3b), and accentuated fluorescence on the
friction trace on the glass surface (Table 3, image
4b). The lipidic vesicles present in the third body at
the beginning of friction burst under the effect of
tribological stresses leading to a lipidic deposit on
the surfaces of the first bodies. This evolution occurs
with a reduction of the friction coefficient from
0.015 to 0.005.
3rd body C (physiological solution, lipidic
bilayers)
As in the case of 3rd body B, the
fluorescence inside the contact before friction was
the same as that observed outside the contact (Table
3, image 1d). But contrary to case B, the
fluorescence did not evolve significantly during
friction (Table 3, image 2d).
Visualisation of the rubbing surfaces of the
first bodies after friction showed that the lipidic
surfaces initially present remained intact (Table 3,
images 3d and 4d).
Correlation of these observations with the
measurement of the very low friction coefficient
(0.0015) from the beginning to the end of friction
showed that the lipidic bilayers adsorbed on the first
bodies resisted friction well and were responsible for
the significant reduction of the friction coefficient in
comparison to 3rd body B.
These results demonstrated that the presence
of two lipidic bilayers separated by a physiological
salt solution layer in the contact area leads to very
low friction. Our very low friction coefficient
values contradict the hypothesis that lipids layers
only reduce wear but not significantly friction in the
healthy joints [6]. Low friction could be due to the
location of velocity accommodation, in the
physiological salt solution layer. This type of
accomodation mode was also suggested by Briscoe
et al. [32], but in their experiments, velocity
accomodation was located in the hydratation water
layer at the substrate/surfactant monolayer interface.
In this study the hydratation layer located between
two lipidic bilayers is probably thicker resulting in a
much lower friction coefficient (by a factor of 10).
3rd body D (substitute synovial fluid, lipidic
bilayers)
The presence of large lipidic vesicles (a few
dozen micrometers in diameter) filled with
hyaluronic acid and albumin gel (synovial gel) in
3rd body D generated uniform fluorescence inside
the contact, but much weaker than outside the
101
contact where it was not uniform (Table 3, image
1f). Therefore the large lipidic vesicles of 3rd body
D did not remain inside the contact, and the
fluorescence observed inside the contact was mainly
caused by the lipidic bilayers adsorbed on the first
bodies.
After 1h of friction, we observed the presence
of fluorescent roller-like structures inside the contact
(Table 3, lower part of image 2f). The roller-like
appearance of this structure could be favoured by the
presence of free synovial gel (not incorporated in
the lipidic vesicles) and, due to the modification of
the velocity accommodation mode, be responsible
for the high friction coefficient (0.12). This value is
similar to those obtained by Benz and Istraelachvili
[33] who studied the friction of a hyaluronic acid gel
fixed (chemically and physically) by a lipidic bilayer
on the surfaces in contact of a surface force
apparatus, and show that synovial gel leads to a high
friction coefficient (0.1 0.3) in a boundary
lubrication regime.
Visualizations of the surfaces of first bodies
after friction show the presence of fluorescent lipidic
vesicles on the HEMA surface (Table 3, image 3f)
and slightly fluorescent non-uniform deposits on the
glass surface (Table 3, image 4f).
Therefore all the experiments carried out on
the artificial cartilage contact model show that the
lipidic bilayers, adhering physicochemically and
uniformly to the rubbing surfaces of the first bodies
led to a very low friction coefficient (in the region of
0.0015). Although the lipidic bilayers adhere during
friction, they do not adhere uniformly (more visible
fluorescent trace of friction in image 4b) which may
explain why the friction coefficient is slightly higher
(0.005). However, a much higher friction coefficient
(0.12) is obtained if the friction is localised inside a
layer of synovial gel and not in the layer of
physiological solution that separates the lipidic
bilayers deposited on the rubbing surfaces.
3.3.2 Steel joint implant contact model
3rd body B (physiological solution, small
lipidic vesicles)
The fluorescence inside the contact before
friction was uniform (Table 4, image 1b) but had
completely disappeared after 1h of friction (Table 4,
image 2b). The small lipidic vesicles (several
hundred nanometres) of 3rd body B were therefore
in the contact before rubbing, in the same way as for
the artificial cartilage contact model.
Visualisation of the friction surfaces of the
first bodies after friction showed that:
the trace of friction on the steel surface did
not contain lipids whereas a uniform bilayer
was observed only outside the friction trace
(Table 4, image 3b),
the lipids adhered to the HEMA surface
uniformly (Table 4, image 4b).
The steel surfaces therefore allowed the
physicochemical adhesion of the lipids, but the
lipidic bilayers did not resist the tribological stress,
explaining the increase in the friction coefficient
from 0.03 to 0.07 from the start to the end of the
tests (Figure 11).
3rd body C (physiological solution, lipidic
bilayers)
As expected, the fluorescence inside the
contact before rubbing was uniform (Table 4, image
1d), due to the lipidic bilayers deposited on the first
bodies.
The fluorescence lost its uniformity during
the test (Table 4, image 2d), since the lipidic bilayers
loosened from the first bodies and accumulated in
fluorescent agglomerations inside the contact.
Visualisations of the surfaces of the first bodies after
friction also showed the destruction of the lipidic
bilayers after friction and the formation of clusters in
the friction traces on the steel (Table 4, image 3d) as
on the HEMA (Table 4, image 4d).
The correlation with a variation from 0.03 to
0.1 of the friction coefficient during the tests (Figure
11) showed that the lipidic bilayers adhering on the
steel surfaces did not resist tribological stress, and
their destruction caused the friction to increase.
3rd body D (substitute synovial fluid, lipidic
bilayers)
The presence of large lipidic vesicles filled
with hyaluronic acid gel and albumin in 3rd body D
generated non uniform fluorescence (Table 4, image
1f), which shows that the vesicles remain in the
contact, as opposed to the case of artificial cartilage.
After 1h friction these vesicles were seen to merge
in the contact and form fluorescent clusters (Table 4,
image 2f).
Furthermore, the images of the surfaces of
the first bodies after friction showed:
the presence of a non uniform fluorescent
deposit on the steel surface over the entire
friction surface (Table 4, images 3e and 3f),
the presence of fluorescent and non
fluorescent clusters at the border of the
contact on the HEMA surface (Table 4,
images 4e and 4f).
These clusters and the deposit could be
caused by the presence of synovial gel not
incorporated in the lipidic vesicles during the
fabrication of the substitute synovial fluid. Velocity
accommodation by shearing of these residues could
be the source of the high friction coefficient (0.12),
as in the case of the artificial cartilage model (Figure
11).
3.3.3 Polyethylene joint implant contact model
3rd body B (physiological solution, small
lipidic vesicles)
The fluorescence inside the contact zone
before friction was uniform (Table 5, image 1b), but
102
lessened substantially after 1h friction (Table 5,
image 2b). The small lipidic vesicles were therefore
present inside the contact before friction and were
mostly ejected from the contact during friction.
The reduction of fluorescence in the contact
zone during the test occurred with the appearance of
fluorescence over the entire surface of the glass
contact, at a level higher than that of the friction
trace (Table 5, image 4b). On the other hand, the
fluorescence of the polyethylene surface remained
negligible (Table 5, image 3b).
This shows that the lipidic vesicles contained
in the volume of 3rd body B and not ejected from
the contact, burst due to tribological stress, with the
lipids adhering only to the glass surface, both
spontaneously and due to the effect of friction.
However, the polyethylene surface did not
permit adhesion by the lipids by spontaneous
physicochemical effects or by tribilogical stress
effects. This appears to result in a high friction
coefficient comparable to that obtained with 3rd
body A (pure physiological solution). The reduction
of the friction coefficient at the end of the test (a
reduction from 0.07 to 0.05) can be explained by the
smoothing of the polyethylene, resulting in increased
shininess of the friction trace (Table 5, image 3a).
3rd body C (physiological solution, lipidic
bilayers)
The presence of lipidic bilayers on the
surface generates fluorescence of the contact (Table
5, image 1d), but this fluorescence is less uniform
than in the case of the articular contact model (Table
3, image 1d) or the steel joint implant contact model
(Table 4, image 1d). This is due to the difference of
wettability between the two surfaces in contact
(glass, polyethylene) which caused faults in the
lipidic bilayers. During friction, the lipids loosened
from the surfaces, to form fluorescent clusters inside
the contact (Table 5, image 2d).
Visualisations of the friction surfaces of the
first bodies after friction showed, as in the case of
3rd body B:
the absence of lipids adhering to the
polyethylene surface (Table 5, image 3d) and
the smoothing of the friction trace detectable
under white light due to the shininess of the
friction trace (Table 5, image 3c).
The presence of a high inhomogeneous
lipidic bilayer on the glass (Table 5, image
4d), with lipidic clusters that did not exist
before friction (Table 2, image 6c).
Thus, in the case of the polyethylene joint
implant contact model, the evolution of the contact
was the same, whether the lipids were initially in the
volume of the 3rd body in the form of small vesicles,
or in bilayers on the surface of the first bodies,
thereby explaining why the change in the friction
coefficient was the same (Figure 11).
3rd body D (substitute synovial fluid, lipidic
bilayers)
The presence of large lipidic bilayers filled
with synovial gel in the third body generated non-
uniform fluorescence of the contact (Table 5, image
1f), which shows that these vesicles exist initially
inside the contact zone. After 1h of friction, fusion
of the vesicles into fluorescent clusters was observed
(Table 5, image 2f).
Visualisations of the friction surfaces of the
first bodies after friction showed:
the absence of fluorescence for the
polyethylene surface (Table 5, image 3f), and
less pronounced shininess in the friction trace
(Table 5, image 3e) than in the two previous
cases, thereby showing the start of surface
smoothing.
the presence of clusters of non-uniform
fluorescence in the contact zone on the glass
surface (Table 5, image 4f).
In this configuration, it therefore seems that
the accommodation of velocity between the surfaces
was mainly ensured by free synovial gel between
fluorescent clusters within the 3rd body, which
explains why the same high level of friction was
found (Figure 11) as for the artificial cartilage and
steel joint implant contacts models. The polythene
was also seen to smoothen during the test, leading to
an apparent reduction of the friction coefficient.
4. CONCLUSIONS
An experimental model was used for the ex-
vivo reproduction of the tribological triplets
associated in, respectively, a healthy joint, a steel
implant, and a polyethylene implant. In particular
the aim was to analyse the tribological role of the
biological components of the natural lubricant
provided by synovial fluid, by specifically focusing
on lipidic structures.
In order to exacerbate the role played by the
interfaces, experimental conditions were chosen so
as to eliminate any hydrodynamic load carrying
capacity effect; therefore third bodies were
considered, making it possible to study the influence
of different lipidic structures on friction separately.
The analysis proposed relied on friction force
measurements associated with optical microscopy
images of the contact and surfaces. This microscopy
made use of fluorescent and white light to detect the
lipidic structures.
The study demonstrated that the presence of
two lipidic bilayers separated by a physiological salt
solution layer in the contact area leads to low
friction, clearly shown in the case of the artificial
cartilage, where it led to a friction coefficient in the
region of a thousandth. On the contrary, it was not
sensitive in the case of the model implants:
103
In the case of the artificial cartilage, the
lipidic bilayers resisted realistic tribological stresses.
In addition, these stresses favoured their formation
in the presence of lipidic vesicles. This beneficial
effect of the lipidic bilayers was not observed in the
implant models.
The steel surfaces also favoured the adhesion
of the lipidic bilayers, which tends to reduce friction.
However, they did not resist tribological stress and
were totally eliminated from the contact after one
hour. Also the friction coefficient returned to a high
value.
The polyethylene surfaces did not permit any
adhesion of the lipidic bilayers. This result appears
to contradict the literature, which states that the
presence of lipids in a steel-polyethylene contact
reduces the friction coefficient [24]. A fall in the
friction coefficient during the tests was observed,
though it is only correlated to the smoothing effect
on the polyethylene surface.
Furthermore, the addition of hyaluronic acid
and albumin in the substitute synovial fluid resulted
in an increase of the friction coefficient in all the
tests performed. This was probably due to the
presence of hyaluronic acid and albumin free gel,
i.e. not incorporated in the lipidic vesicles. It was
proposed to explain the increase of the friction
coefficient that large molecules of hyaluronic acid
may bridge the gap between surfaces [6] even in the
presence of lipidic bilayers [34]. Within the
framework of this study, we correlated the increase
of the friction coefficient to the formation of the
rollers containing lipids, which have been visualized
in fluorescence microscopy (Table 3, image 2f).
However, it is probable that the presence of
free synovial gel was due to our method of preparing
the initial synovial gel - lipidic vesicle solution,
which did not permit the complete incorporation of
the gel. Indeed, it appears that the process of
incorporating the gel in the vesicles greatly depends
on physicochemical conditions (temperature, pH,
osmotic pressure, etc.) [19]. Also, the
pharmaceutical synthesis of the liposomes includes a
final filtration step aimed at eliminating the free gel.
This filtration was not performed in the framework
of this work.
What is more, it is probable that there is no
free gel in a healthy joint, in which the presence of
lubricin ensures the correct formation of lipidic
structures (bilayers, vesicles), by providing an
adhesive interface between the lipidic surfaces of the
cartilage and the synovial gel.
The situation is perhaps quite different in
implants. This study suggests that the lipidic bilayers
are destroyed by friction, thereby permitting the
existence of free gel and increasing friction.
Nonetheless, it should be noted that in spite of an
increase in the friction coefficient, the presence of
free gel may have a beneficial effect in protecting
the steel surfaces against wear [1].
This study therefore shows that molecular
structures, such as lipidic bilayers, hyaluronic acid
and albumin gel, and theirs interactions have a
decisive influence on the tribological performances
of the artificial cartilage and the joint implant
materials operating under boundary regime. Thus it
appears vital to take them into account in implant
lifetime tests and not use a lubricant composed only
of physiological solution and albumin, as is most
often the case.
A means of optimising the rubbing surfaces
of joint implants requires improving the
compatibility of the materials with the lipidic
bilayers, in order to obtain a low friction coefficient
and a lubrication mode similar to that of a healthy
joint. These conclusions agree with the works of
Hills [25] which show a stack of 3 to 7 lipidic
bilayers on the surfaces of articular cartilage and
suggest that most implant surfaces do not permit the
formation of this stacking.
Our results with lipidic multilayers show very
low friction coefficients similar to those obtained for
polyelectrolyte experiments [Klein]; thus future
experiments should be done in order to properly
identify the tribological role of these molecules.
Also, it was shown that polyelectrolytes such as
lubricin have a role of adhesion on the lipidic
membranes ([34,13]) which could modify their
tribological performance. Therefore, future
experiments should be performed in presence of
lubricin and lipidic multilayers.
ACKNOWLEDGEMENT
The authors would like to thank in particular:
Mr. G. Vitally of CORNEAL Industrie which
provided the HEMA samples,
Pr. L. Cheze, Pr. J.-P. Carret of the
Laboratoire de Biomcanique et Modlisation
Humaine (LBMH) of Universit Claude Bernard,
Lyon 1 for their help in providing understanding of
joint dynamics, anatomy and articular histology.
Pr. D. Hartmann of the Institut des Sciences
Pharmaceutique et Biologique de Lyon, UMR MA,
for his help in providing understanding of the
biochemistry and morphology of articular molecular
structures.
M. C. Godeau for his helpful participation in
this work.
REFERENCES
1. Sawae Y., Murakami T., 2006,An
Experimental Investigation of Boundary Lubrication
Mechanism with Protein and Lipid in Synovial Joint
Using Total Internal Reflection Fluorescence
Microscopy, Journal of Biomechanics; Vol. 39
(Suppl 1).
104
2. Wimmer MA, Sprecher C, Hauert R, Tger
G, Fischer A. 2003, Tribochemical Reaction on
Metal-on-Metal Hip Joint Bearings. A Comparison
Between In-Vitro and In-Vivo Results, Wear; 255 :
10071014.
3. Schurz J, Ribitsch V., 1987, Rheology of
synovial fluid, Biorheology, 24(4): 385-99
4. Dowson, D., and Jin, Z.M., 1987, An Analysis
of Micro-Elastohydrodynamic Lubrication in
Synovial Joints Considering Cyclic Loading and
Entraining Velocities, Fluid Film Lubrication
Osborn Reynolds Centenary, Proc. 23th LeedsLyon
Symposium on Tribology, edited by D. Dowson et
al., Elsevier, Amsterdam, 1986, pp. 375386.
5. Swann D.A., Silver F.H., Slayter H.S., Stafford
W., Shore E., 1985, The Molecular Structure and
Lubricating Activity of Lubricin Isolated from
Bovine and Human Synovial Fluids, Biochem J;
225: 195-201.
6. Jay G.D., Harris D.A. and Cha C.-J., 2001,
Boundary Lubrication by Lubricin is Mediated by
O-linked (1-3)Gal-GalNAc Oligosaccharides,
Glycoconjugate Journal 18, 807815
7. J. Klein, 2006, Molecular Mechanisms of
Synovial Joint Lubrication, Proc. IMechE Vol. 220
Part J: J. Engineering Tribology, 220, 691-710.
8. Schwarz IM, Hills BA., 1998, Surface-Active
Phospholipid as the Lubricating Component of
Lubricin, British Journal of Rheumatology; 37: 21-
26.
9. Hills BA., 1989, Oligolamellar Lubrication of
Joints by Surface-Active Phospholipid. J
Rheumatol;16: 82-91.
10. Oates KMN, Krause WE, Jones RL and Colby
RH., 2005, Rheopexy of Synovial Fluid and
Protein Aggregation. Journal of the Royal Society
Interface; 1-8.
11. Pasquali-Ronchetti, 1997, Hyaluronan
Phospholipid Interactions, Journal of structural
biology; 120: 110.
12. Crescenzia V, Taglienti A, Pasquali-Ronchetti
I., 2004, Supramolecular Structures Prevailing in
Aqueous Hyaluronic Acid and Phospholipid
Vesicles Mixtures: An Electron Microscopy and
Rheometric Study, Colloids and Surfaces A:
Physicochem. Eng. Aspects; 245: 133135.
13. Rhee DK, Marcelino J, Baker MA, Gong Y,
Smits P, Lefebvre V, Jay GD, Stewart M, Wang
H, Warman ML, Carpten JD., 2005, The
Secreted Glycoprotein Lubricin Protects Cartilage
Surfaces and Inhibits Synovial Cell Overgrowth of
Synovial Cell Growth, The Journal of Clinical
Investigation; 115(3): 622631.
14. Schvartz I, Seger D, Shaltiel S., 1999,
Molecules in Focus: Vitronectin, The
International Journal of Biochemistry & Cell
Biology; 31: 539-544.
15. Tolosano E, Altruda F., 2002, Hemopexin:
Structure, Function, and Regulation, DNA and Cell
Biology; 21(4): 297-306.
16. Trunfio-Sfarghiu AM, Berthier Y, Meurisse
MH, Rieu JP. Operation Function of a Healthy
Synovial Joint. Part I: Design of a Tribological
Model by Multiscale Analysis of the Role of
Biological Components, submitted for publication.
17. Broom ND, Oloyede A., 1998, The Importance
of Physicochemical Swelling in Cartilage Illustrated
with a Model Hydrogel System, Biomaterials; 19:
1179-1188.
18. Freeman ME, Furey MJ, Love BJ, Hampton
JM., 2000, Friction, Wear, and Lubrication of
Hydrogels as Synthetic Articular Cartilage, Wear;
241: 129-135.
19. Covert RJ, Ott RD, Ku DN., 2003, Friction
Characteristics of a Potential Articular Cartilage
Biomaterial, Wear; 255: 10641068.
20. Dowson, D., and Jin, Z. M., 1992,
Microelastohydrodynamic Lubrication of Low-
Elastic-Modulus Solids on Rigid Substrates, J.
Phys. D, 25, pp. A116A123.
21. K. Nakashima, Y. Sawae, T. Murakami, 2005,
Study on Wear Reduction Mechanisms of Artificial
Cartilage by Synergic Protein Boundary Film
Formation, JSME International Jurnal, series C,
vol 48, nr. 4, p 555-561
22. Migliaresi C, Nicodemo L, Nicolais L,
Passerini P., 1981, Physical Caracterisation of
Microporosus Poly (2-Hydroxyethyl Methacrylat)
Gels, Journal of Biomedical Materials Research;
15: 307-317.
23. Mow VC, Ratcliffe A., 1992, Cartilage and
diarthrodial joints as paradigms for hierarchical
materials and structures. Biomaterials; 13(2): 67-
97.
24. Watanabe M, Leng CG, Toriumi H, Hamada
Y, Akamatsu N, Ohno S., 2000, Ultrastructural
Study of Upper Surface Layer in Rat Articular
Cartilage By In-Vivo Cryotechnique Combined
with Variosus Tratements, The Clinical Electron
Microscopy Society of Japan; 33: 16-24.
25. Torchillin V, Weissig V., 2003, Liposomes: A
Practical Approach, Oxford University Press, USA,.
26. He L, Dexter A, Middelberg A., 2006,
Biomolecular Engineering at Interfaces, Chemical
Engineering Science; 61: 989 1003.
27. Bayerl TM, Bloom M., 1990, Physical
Properties of Single Phospholipids Bilayers
Adsorbed to Micro Glass Beads: A New Vesicular
Model System Studied by 2H-Nuclear Magnetic
Resonance, Biophys. J.; 58(2): 357362.
28. Hills BA, Crawford RW., 2003, Normal and
Prosthetic Synovial Joints Are Lubricated by
Surface-Active Phospholipid. A Hypothesis, The
Journal of Arthroplasty; 18( 4): 499-505.
29. see the Olyumpus microscopy resource center,
http://www.olympusmicro.com/primer/anatomy/refl
ectkohler.html
105
30. Gale LR, Coller R, Hargreaves DJ, Hills BA,
Crawford R., 2006, The Role of SAPL as a
Boundary Lubricant in Prosthetic Joints. Tribology
International; Available online 18 January.
31. Perie D., Hobatho M.C., 1998, In Vivo
Determination of Contact Areas and Pressure of the
Femorotibial Joint Using Non-Linear Finite Element
Analysis, Clinical Biomechanics 13, 394-402
32. Johnson T.S., Laurent M.P., Yao J.Q.,
Blanchard C.R., 2003, Comparison of Wear of
Mobile and Fixed Bearing Knees Tested in a Knee
Simulator, Wear 255, 11071112
33. Briscoe et al., 2006, Boundary Lubrication
Under Water, Nature 444, 191-194.
34. Benz M, Chen N, Istraelachvili J., 2004,
Lubrication and Wear Proprietes of Grafted
Polyelectrolytes, Hyaluronan and Hyalan, Mesured
in the Surface Forces Apparatus. Journal of
Biomedical Materials Research Part A; 71A(1): 6-
15.
35. Bruno Zappone, Marina Ruths, George W.
Greene, Gregory D. Jay, Jacob N. Israelachvili,
Adsorption, Lubrication and Wear of Lubricin on
Model Surfaces: Polymer Brush-Like Behavior of a
Glycoprotein, Biophysical Journal, 2006
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 106-112
D. N. OLARU
e-mail: dolaru@mail.tuiasi.ro
C. STAMATE
A. DUMITRASCU
Gh. PRISACARU
Department of Machine Elements and
Mechatronics,
Technical University Gheorghe Asachi Iasi
ROMANIA
ROLLING FRICTION TORQUE IN
MICROSYSTEMS
To determine the rolling friction torque in the micro rolling
systems, the authors developed an analytical model based on the
dissipation of the inertial energy of a rotating microdisc in three
rolling microballs. Using an original microtribometer with two
steel rotating discs and three steel micro balls the rolling friction
torque in dry conditions was determined for contacts loaded with
normal forces of 8.68 mN to 33.2 mN and with rotational speed
ranging between 30 to 210 rpm. The experimental results confirm
the hypothesis that the rolling friction torque in dry contacts is not
depending of the rotational speed.
Keywords: rolling friction torque, microtribometer, dynamic
modeling
1. INTRODUCTION
The use of the rotating microball bearings in
MEMS applications (micromotors, microgenerators,
microactuators, micropumps) implies the
simplification in construction, low level friction, low
level wear, high stability. Thus the microball
bearings seem to be a promising solution for future
MEMS applications.
Recently, some experimental evaluations of
the global friction in the rotating microball bearings
was realized. Ghalichechian et al. [1] determined
experimentally the global friction torque in an
encapsulated rotary microball bearing mechanism
using silicon micro fabrication and stainless steel
microballs of 0.285 mm diameter. The global
friction torque was indirectly obtained by measuring
the transient response of the rotor in the deceleration
process from a constant angular velocity until it
completely stops due to friction. Using a high
speed camera system, the angular position of the
rotor in the deceleration process was determined.
The authors introduced the hypothesis that the global
friction torque in the microball bearing is
proportional with rotational speed. In this
circumstances, the measured angular positions
(t) | was fitted to an exponential function of the
form
b t
(t) a e c t d

| = + + , where t is the time (in

seconds) and a, b, c and d are constants. The
acceleration of the rotor was obtained by
differentiation of the function (t) | . The global
friction torque obtained varied between 5.62 Nm
and 0.22 Nm for the rotational speed of the rotor
decreasing from 20.5 rad/s to zero, under an axial
load of 48 mN. McCarthy et al. [2] experimentally
investigated the influence of the speed and of the
normal load on the friction torque in a planar-
contact encapsulated microball bearing having 0.285
mm diameter steel balls and silicon races. Using the
spin-down testing and the hypothesis of the linear
dependence between global friction torque and
rotational speed, the authors determined the global
friction torque for rotational speed between 250 rpm
and 5000 rpm and for axial loads between 10 mN
and 50 mN. Based on the experimental results, the
authors obtained following empirical power-law
model for the global friction torque in the microball
bearing [2]:
5 0.444
N
M 9 10 F n

= , where M is the
global friction torque in Nm, F
N
is the axial load
acting on the microball rolling, in mN, and n is
rotational speed in rot/min. Tan et al. [3] propose a
viscoelastic model for friction force developed in a
rolling contact between a microball and a plane.
This viscoelastic model includes material
parameters, ball diameter, normal load and linear
speed and was applied for a steel microball having
0.285 mm diameter, loaded with a normal force of 2
mN and rolling on a silicon plane with a linear speed
between zero to 0.03 m/s. The rolling friction torque
obtained by this viscoelastic model varied between
zero and
3
2.2 10

Nm.
Using the integration of the free oscillations
equations of a steel microball on a spherical glass
surface, Olaru et al. [4] evaluated the rolling friction
torque on the basis of the number and amplitude of
the experimentally determined microball
oscillations. For a steel microball having a diameter
of 1 mm, Olaru et al. [4] obtained in dry conditions
values for rolling friction torque of
3
0.7 10

Nm at
a normal load on microball of 0.04mN.
107
The experimental results obtained in [1] and
[2] reefer to the global rolling friction torque in a
rotary microball bearing. It is important to notice
that in a rotary microball bearing the global friction
torque is a result of both rolling and sliding friction
caused by the pivoting motion of the microballs and
by the direct contact of the microballs.
To determine only the rolling friction torque
in the micro rolling systems the authors developed
an analytical model based on the dissipation of the
inertial energy of a rotating microdisc in three
rolling microballs. Using an original
microtribometer with two steel rotating discs and
three steel microballs, the rolling friction torque in
dry conditions was determined for normal contact
loads of 8.68 mN to 33.2 mN and for rotational
speed varying between 30 to 210 rpm.
2. ANALYTICAL MODEL
Figure 1 presents the new micro tribometer.
The driving disc 1 is rotated with a constant
rotational speed and has a radial groove race. Three
microballs are in contact with the race of the disc 1
at the equidistance position (120 degrees). All three
microballs sustain an inertial disc 2 and are normal
loaded with a force
G
Q
3
= , where G is the weigh of
the disc 2. When the disc 1 start to rotate with a
constant angular speed
1
, the balls start to rolls on
the raceway of the disc 1 and start to rotate the
inertial disc 2, as a result of rolling friction forces
between the balls and the disc 2. As a result of
inertial effect the disc 2 is accelerated from zero to
the synchronism rotational speed (when
2
=
1
) in
a time t, after that the rotational speed of the disc 1 is
stopped. The disc 2 has a deceleration process from
the constant rotational speed
2,0
to his completely
stop as a result of the friction in the rolling of the
three microballs over the two discs.
In the deceleration process of the disc 2 when

2
decreases from a constant value to zero,
following differential equation can be used:
2
2 f
d
J 3 F r M 0
dt
e
= , (1)
where J is inertial moment for the disc 2, F
2
is the
tangential force developed in the contact between a
microball and disc 2, r is the radius and M
f
is the
friction torque developed between the rotating disc 2
and air.
For a disc with inner radius R
i
, outer radius
R
e
and a mass m
d
, the inertial moment J is
determined by following relation:
2 2
d i e
J 0.5 m (R R ) = + . (2)
For a disc having a rotational speed
2
e in a
fluid with a kinematics viscosity
f
u and a density
f
, the friction torque M
f
can be determined by
relation [5]:
5 2
f M f
M 0.5 K R = e , (3)
where K
M
is a coefficient depending on the
Reynolds parameter. When the rotational speed of
the disc 2 have maximum values between 30 and
210 rpm and the radius R is 0.012 m, the Reynolds
parameter have values between 30 and 200 (the
kinematics viscosity of the air was considered
6
f
15 10

u = m
2
/s and the density of the air was
considered
f
= 1.18 kg/m
3
). For these values of the
Reynolds parameter it can be approximated the K
M
coefficient by a constant value of 0.5 and equation
(3) can be approximated as follows:
2
f f
M c = e , (4)
where the coefficient c
f
have an approximate value
of
11 2
7.3 10 N m s

.
Figure 1. General view of the microtribometer
108
As presented in Figure 2, in the deceleration
of the disc 2, following forces act on a microball in
the rotational plane: the tangential contact forces F
1
and F
2
and the inertial force F
ib
. Also, in the two
contacts we consider two rolling friction torques M
r1
and M
r2
.
Figure 2. The forces and the moments acting on a
microball in deceleration process
The tangential force F
2
was determined using
force and moment equilibrium equations for a
microball, resulting:
ib r1 r 2
2
F (M M )
F
d 2
+
= , (5)
where d is the microball diameter.
The inertial force acting in the center of the
microball is determined by relation:
b
ib b
d
F m r
dt
e
= , (6)
where m
b
is the mass of the microball and
b
e is the
angular speed of the microball in the revolution
motion around the center of the two discs.
Considering the pure rolling motion of the
microballs, the angular speed
b
e can be expressed
as
b 2
0.5 e = e and the equation (6) can be written
as follows:
b 2
ib
m r d
F
2 dt
e
= . (7)
According to the equations (1), (4), (5) and
(7), following differential equation in the
deceleration process of the disc 2 is obtained:
2 2
r1 r 2 2
d
a (M M ) b
dt
e
= + + e , (8)
where a and b are constants defined by:
2
b
3 r
a
3
d (J r m )
4

=
+
,
f
2
b
c
b
3
(J r m )
4
=
+
.
To integrate the differential equation (8), two
hypotheses were made:
i) it is considered that the rolling friction
torques Mr1 and Mr2 are not depending on the
rotational speed;
ii) the rolling friction torques Mr1 and Mr2
have a linear dependence on rotational speed.
i) Considering that the rolling friction torques
M
r1
and M
r2
are constants, equation (8) leads to the
following solution for
2
as function of time:
2 2,0
c b
(t) tg c t arctg
b c
1 | |
e = + e
| (
\ . ]
, (9)
where
r1 r2
c a b (M M ) = + and
2,0
e is angular
rotational speed of the disc 2 at the moment of the
stopped the rotation of the disc 1.
Considering that
2
2
d (t)
(t)
dt
|
e = , where
2
(t) | is the variation of the angular position of the
disc 2 in deceleration process, equation (9) can be
integrated and following solution for
2
(t) | results:
2,0
2
2
2,0
b
ln 1 tg c t arctg
c
(t)
2b
b
ln 1
c
.
2b
1 1 | |
+ + e
( | (
\ . ] ]
| = +
1
| |
+ e (
|
\ .
(
]
+
(10)
ii) Considering that the rolling friction
torques M
r1
and M
r2
have a linear dependence on
rotational speed it can be written that
r1 r 2 2
(M M ) k + = e and differential equation (8)
becomes:
2 2
2 2
d
a k b
dt
e
= e + e . (11)
Equation (11) leads to the following
solutions:
2
a k exp( a k t k1)
(t)
1 b exp( a k t k1)
+
e =
+
; (12)
2
1
(t) ln(1 b exp( a k t k1))
b
1
ln(1 b exp(k1)).
b
| = +

(13)
where
2,0
2,0
k1 ln
a k b
| | e
= |
|
+ e
\ .
.
For given dimensions of the microballs and
of the two discs, by monitoring the angular position
109
and angular speed of the disc 2 in the deceleration
process it is possible to determine the sum of the
rolling friction torques
r1 r2
(M M ) + .
Also, having determined the sum of these
friction torques, the tangential force F
2
can be
determined by equation (5) and the friction
coefficient in the rolling contact
r
by equation:
2
r
F
Q
= . (14)
3. EXPERIMENTAL INVESTIGATION
Using the new microtribometer presented in
Figure 1 a set of experimental investigations was
performed. The microtribometer was mounted on
the rotational table of the CETR-UMT Tribometer
as shown in Figure 3.
To determine the angular acceleration of the
disc 2 a high speed camera Philips SPC900NC/00
VGA CCD with 90 frames/seconds was used to
capture the angular position of the disc 2 from the
rotational speed
2,0
e to his completely stop. Also,
the angular positions of the disc 1 are captured by
camera. In Figure 4, the registered positions of the
disc 2, and of the disc 1, at a short time t after the
stop of the disc 1 are presented.
The images captured by the camera was
processed, frame by frame, in a PC using Virtual
Dub soft and was transferred in AutoCAD to
measure the angular positions
2
corresponding to
every frame. The camera was installed vertically
150 mm above the disc 2, to minimize the
measurement errors. A white mark was placed both
on disc 2 and on disc 1 as it can be observed in the
Figure 4 and the angular positions
2
(t) was
measured according to the reference position of the
mark on the disc 1(position at t = 0). The discs 1
and 2 are the steel rings of an axial ball bearing
(series 51100) having a rolling path at a radius r =
8.4mm and a transversal curvature radius of 2.63
mm. The inertial disc 2 was machined on external
surface by electro erosion to reduce the weight to a
minimum of G = 26.05 mN, and it has the following
dimensions: R
i
= 5 mm, R
e
= 12 mm, which means a
minimum normal load on every microball Q = 8.68
mN. To increase the normal load on the microball a
lot of new discs similar to the disc 2 was attached on
the disc 2 obtaining following values for the normal
load: 8.68 mN, 15 mN, 22.3 mN, 27 mN, 33.2 mN.
Three stainless steel microballs having the diameter
of 1.588 mm (1/16 inch ) was used in the
experiments. The roughness of the active surfaces
of the two discs and of the balls was measured by
Form Talysurf Intra System. Following values of
Ra were obtained: rolling path of the disc 1 and 2,
Ra = 0.030 m, and ball surface, Ra = 0.02 m. The
tests were realized for the following rotational speed
of the disc 2: 30 rpm, 60 rpm, 90 rpm, 120 rpm, 150
rpm, 180 rpm, 210 rpm.
All measurements are performed in steady
room environment at a temperature of (18-20)
0
C
and a relative humidity of (40 50)%RH. All the
tests were realized in dry conditions (without
lubricant or condensed water on contact surfaces).
Figure 3. General view of the experimental equipments
110
Figure 4. Determination of the angular position
2
(t) of the disc 2
4. VALIDATION OF THE ANALYTICAL
MODELS
Two experimental data were obtained for
each experiment: the variation of the angular
position
2
(t) from the moment of beginning the
deceleration process of the disc 2 to its completely
stop and the time of the deceleration process. A
typical variation of the angular position
2
(t) for a
rotational speed of 120 rpm and a normal load Q =
8.68 mN experimentally determined, is shown in
Figures 5 and 6. For all experiments, the variations
of the angular position of the disc 2,
2
(t) are
similar, but other time of deceleration and other
maximum values were obtained, depending of the
initial angular speed
2,0
e and the normal load Q
acting on the microballs. Both hypotheses were
used to validate the experimental results.
i) The hypothesis of constant torque friction
was applied for all experiments and a good
validation with experiments was obtained. Using
equation (10) it was determined the value of the sum
of friction torques (M
r1
+ M
r2
) imposing the
condition that at the stop of the disc 2, the angular
position of this disc cumulates the experimentally
determined value. With the above sum (M
r1
+ M
r2
),
it was verified by equation (9) if the angular speed
of the disc 2 was stopped after the experimentally
determined time.
Figure 5 shows the numerical variation of the
angular position of the disc 2 given by equation (10)
for a rotational speed of the disc 2 of 120 rpm and a
normal load Q =8.68 mN.
The maximum differences between the
numerical values obtained by equation (10) and the
experimental values do not exceed 5%. In Figure 5-
b it can be observed the numerical variation of the
angular speed
2
(t) e obtained by equation (9) with
a quasi linear variation from
2,0
e =12.4 rad/s to
2,0
e =0, in a time t = 41 seconds. This deceleration
time corresponds to the experimental determined
value.
ii) The hypothesis of the linear variation of
the friction torque with rolling speed was applied for
all experiments. Using equation (13), the value for
the sum of friction torques (M
r1
+ M
r2
) was
determined by imposing the condition that at the
stop of the disc 2 the angular position of this disc
cumulates the experimentally determined value.
With the above sum (M
r1
+ M
r2
) equation (12) yields
the variation of the angular speed of the disc 2.
(a)
(b)
Figure 5. Variation of the numerical and
experimental values for
2
(t) (a) and variation of
the angular speed
2
(t) e -(b) for a rotational speed
of the disc 2 of 120 rpm and normal load Q =8.68
mN (constant friction torques hypothesis)
111
Figure 6 shows the numerical variation of the
angular position of the disc 2 given by equation (13)
for a rotational speed of the disc 2 of 120 rpm and a
normal load Q =8.68 mN. The maximum
differences between the numerical values obtained
by equation (13) and the experimental values do not
exceed 8%. Figure 6-b shows the numerical
variation of the angular speed
2
(t) e obtained by
equation (12). It can be seen that the angular speed
of the disc 2 is not zero at a time t = 41 seconds.
(a)
(b)
Figure 6. Variation of the numerical and
experimental values for
2
(t) (a) and variation of
the angular speed
2
(t) e -(b) for a rotational speed
of the disc 2 of 120 rpm and normal load
Q =8.68mN (variation of friction torques
hypothesis)
By comparing the two analytical variations of
the angular position
2
(t) given by equations (10)
and (13) and presented in Figure 7, it can be
observed that the equation (10) leads to a variation
of angular position with a maximum around of the
time t = 41 seconds while the equation (13) leads to
a continuum increasing of the angular position
2
(t).
This means theoretically a continuous rotation over
the time of experimentally stopping of the disc 2 .
Figure 7. The theoretical variation of the angular
position
2
(t) for the two hypothesis applied for
rotational speed of 120 rpm and normal load of
8.68 mN
Our conclusion is that the hypothesis of the
constant friction torque can be accepted and leads to
a good theoretical model in the interval of the
rotational speed between 30 rpm to 210 rpm.
5. EXPERIMENTAL RESULTS
The sum of the friction torques for all
experiments was determined in the hypothesis of the
constant friction torque using equations (9) and (10).
Considering that the geometry of the contact
between microball and the two discs is the same and
neglecting the influence of the microball weight (the
mass of a microball leads to an additional force Q
b
= 0.165mN in the contact between microball and the
disc 1) we can consider that the friction torque
between a microball and the disc 1 or 2 is given by
relation M
r
= 0.5(M
r1
+ M
r2
). In Figure 8 are
presented the rolling friction torques M for all
rotational speeds and normal loads used in the
experiments.
It can be observed that between 30 rpm and
150 rpm the friction torque M depends only on the
normal load and is not depending on the speed. By
increasing the speed from 150 rpm to 210 rpm the
friction torque increases with rotational speed,
especially when increasing the normal load. The
increasing of the friction torque with rotational
speed above 150 rpm, can be explained by
increasing of the rotating discs vibration with a
supplementary loss of energy. It is important to
notice that increasing of the normal load is realised
by adding supplementary discs on the initial disc 2.
Geometrical imperfections of the supplementary
discs increase the vibration level of the rotating disc,
and vibrations of the rotating disc were observed
experimentally.
112
Figure 8. The rolling friction torques M
r
determined by equation (10) applied to the experiments
The friction coefficient determined by
equation (14) has values between 0.0002 and 0.0004
which means a dominance of the rolling friction
between the microballs and the two discs.
6. CONCLUSIONS
Two analytical models to determine the
rolling friction in an original microtribometer were
elaborated. The two models are based on the
integration of the differential equation of a rotating
disc sustained only by three microballs. Two
hypothesis were considered : i) the friction torque is
not depending on the rotational speed in dry contacts
and ii) the friction torque has a linear variation with
rotational speed.
To validate these hypotheses, a set of
experiments was performed for a variation of
rotational speed between 30 rpm to 210 rpm and a
normal load in the rolling contact between 8.68 mN
to 33.2 mN. The hypothesis based on the constant
friction coefficient was validated as a good
hypothesis in dry conditions.
The friction torques for all experiments was
determined by the analytical model based of the
constant friction torque. The numerical values are
between 1.8 N.mm to 7.2 N.mm .
The rolling friction coefficient obtained in all
experiments ranges between 0.0002 and 0.0004.
ACKNOWLEDGEMENTS
This paper was realised with the support of
Grant CNCSIS ID_607 No. 381/1.10.2007 and
BRAIN Doctoral scholarships as an investment in
intelligence project, financed by the European
Social Found and Romanian Government.
REFERENCES
1. Ghalichechian, N., Modafe, A., Beyaz, M. I.,
Ghodssi, R., 2008, Design, Fabrication, and
Characterization of a Rotary Micromotor Supported
on Microball Bearings, Journal of
Microelectromechanical Systems, 17, p. 632-642
2. McCarthy, M., Waits, C. M., Ghodssi, R.,
2009, Dynamic Friction and Wear in a Planar-
Contact Encapsulated Microball Bearing Using an
Integrated Microturbine,. Journal of
Microelectromechanical Systems, 18, p. 263-273
3. Tan, X., Modafe, A., Ghodssi, R., 2006,
Measurement and Modeling of Dynamic Rolling
Friction in Linear Microball Bearings, Journal of
Dynamic Systems, Measurement, and Control, 128,
p. 891-898
4. Olaru, D. N., Stamate, C., Prisacaru, Gh.,
2009, Rolling Friction in a Microtribosystem,
Tribology Letters, 35, p. 205-210
5. Czichos, H. ed. HTTE - Die Grundlagen der
Ingenieurwissenschaften, Springer Verlag, Berlin,
1989.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 113-119





Lorena DELEANU
e-mail: lorena.deleanu@ugal.ro

Sorin CIORTAN


Machine Design and Graphics,
University Dunarea de Jos Galati,
ROMANIA

EVALUATING TRIBOLOGICAL DAMAGES BY
3D PROFILOMETRY
The authors present a study on using 3D roughness parameters for
assessing the quality of worn surfaces of polymeric composites. A
set of three plates was tested under water lubrication in contact with
a steel disc, being tested at 2.5m/s (the sliding speed at plate center)
and average pressure 2.02 MPa. The plates (6 x 20 x 30 mm) were
made of PTFE composites with glass fibers (0% for the polymer,
15%, 25% and 40%, respectively).
Keywords: polymeric composites, wear, roughness


1. INTRODUCTION
Polymeric composites are expected to give
solutions for tribological applications as
manufacturing technology and test results offer the
opportunity of an easy adapting to the design
requirements. PTFE composites are still used even if
there are some other fluoropolymers trying to
challenge it, as this polymer offers, especially in
composites, the possibility of friction reduction. The
new tendency is to use PTFE as adding material as
solid lubricant and less as matrix, but there are
several applications including those requiring
chemical resistance in which PTFE could be an
efficient matrix [2,5,6,8] or an adding material [3,7]
for improving tribological behavior.
This paper investigated the surface quality of
triboelements made of PTFE and PTFE + glass
fibers when sliding in water against steel in order to
evaluate 3D roughness parameters and to point out
correlations among the analyzed parameters and the
constituent percentage.
2. MATERIALS AND TESTING
METHODOLOGY
Tests were done on four materials and their
mechanical properties are given in Figure 1. Testing
machine has an original design in order to allow a
large range for sliding speed but low variations
(5%) and loading (010 kN3%) (Fig. 2, [2]).
Each test involved a set of three plates (6x20x30
mm), introduced in a steel support disc. The mating
disc was made of stainless steel (40 HRC and
Ra=0.60.8m). Plates were made of PTFE and
PTFE composite with different glass fiber
concentration. Testing conditions were: sliding
speed v=0.7, 1.5, 2.5 m/s, values for average
pressure being p=0.22, 0.77, 1.46 and 2.02 MPa,
respectively, open-circuit water temperature
=181C. Average pressure was calculated as
plate
F
p
3A
[MPa], (1)
where F is the normal load [N], n=3 is the number of
tested plates in one set and A
plate
the nominal area
of one plate [mm
2
].
Friction coefficient was calculated based on
the outputs from the torsion gauge 6,
f r
F F
F M r

a
, (2)
where F is the normal load, F
f
the friction force,
M
r
the resistant torque as measured by the gauge 6
(fig. 2) and r
a
the radius from the rotation axle of
the steel disc to the center of the plate, during a
sliding distance of 5,000 m (rate sampling being
1/sec). Plates position may be changed in order to
obtain different sliding speed, allowing also to
calculate average wear if r
a1
=r
a2
=r
a3
.


0
20
40
60
80
PTFE PTFE +
15% GF
PTFE +
25% GF
PTFE +
40% GF
Traction limit (MPa)
Shore hardness


Figure 1. Tested materials (gf glass fibers)
114
Figure 2. Testing device and samples placement:
1 - driving shaft, 2 - enclosure,
3 - mobile triboelement, 4 - fixed triboelement (with
plates 1, 2 and 3), 5 - base board, 6 - torsion gauge
(a)
(b)
Figure 3. a) investigated surfaces on the same
sample, for 5 measurements; b) SEM image
of B1 sample
For each material and for the testing
conditions (v=2.5 m/s, p=2.02 MPa), a plate from
each set of three, was the subject of this
investigation, using a CETR contact profilometer
and its dedicated soft for analysis [9]. There were
recorded the topography of 5 zones of 500 m x m
in the central region of the plate, one next to the
other, 3 in the sliding direction and 2 in the radial
direction, coded as in Figure 3 [10].
Figure 3 presents the investigated zones for
the composite with 40% GF. The results of the
analyzed parameters for one of the three plates that
forms a tested set for each of the studied materials
are given in Figures 5 and 6.
A reduced number of measurements could
induce evaluation errors as 3D investigations the
authors having access to, are small (500 m x 500
m), especially on Sq and Ssk as they point out local
topography disturbance. Sa seems to be unaffected
by the number of measurements and also Sk, but
Sku, Svk have high maximum values above the
obtained average. The only parameters having the
spread of values around 16% as recommended by
[10], are Sa and Sk.
3. RESULTS AND DISCUSSIONS
3.1. Tribological behavior
Wear as average mass loss of a plate after
5,000 m of sliding in water in open circuit, is given
in Figure 4a for the test regime and Figure 4b
presents the evolution of friction coefficient for the
same regime. In this paper quality investigation is
done only for the sliding regime characterized by an
average pressure of p=2.02 MPa and a sliding speed
of v=2.5 m/s. Negative values for wear are possible
because of continuous process of fragmenting and
embedding of wear debris together with small water
droplets and water solid impurities that remain
insulated into the superficial layers, increasing the
plate mass.
SEM images in Figure 6 reveal mechanical
processes characterizing the superficial layers of the
polymer and composites and they were done after
testing under the conditions (v=2.5 m/s, p=2.02
MPa, water lubrication in open circuit):
the polymer has a different behavior when
tested in contact with steel counterpart, including
abrasion, localized flows, transfer on the steel
surface, material detaching as rolled particles, re-
embedding of wear particles etc.,
for composites the processes differ in
intensity and aspects: the polymeric matrix has
lower displacements and reveal neither deep plough
traces, nor overlapping. the random fiber net allows
reducing the polymer flow and detaching, but glass
fibers on the surface are bearing enough load to be
worn, fractured at medium glass fiber concentration
(1525%wt); statistically, fibers remained on the
115
surface have been fractured at their end situated
on/out the surface, the fragments being embedded
near-by and, thus, consolidating the polymer in the
fiber neighborhoods, but at higher concentration
(40% GF), many fibers are totally fractured, around
the length middle, the process of fiber agglomeration
being the result of wearing (tearing) out the polymer,
the external load being now supported by a rigid
structure, formed by the random arrangement of
fibers within the superficial layer, a similar process
being analyzed in [4, 5].
For the tested composites, at v=2.5 m/s, the
friction coefficient becomes stable for all the
composites, except for the polymer that varies
within the range 0.0080.02; there is a general
tendency that friction coefficient has greater values
at starting, but it becomes stable after ~2,500 m of
sliding (Fig. 4b).
By analyzing the wear diagram (Fig. 4a), the
following aspects may be pointed out:
composites have a better tribological
behavior as their wear is four, even ten times or less
than the polymer wear, under similar testing
conditions;
for the set of three plates involved in each
test, results are spread in a large range (1015%
around average value, calculated as
m=(m
1
+m
2
+m
3
)/3, the spread being larger for
the polymer and the composite with the highest
concentration, 40% GF);
comparing only these four tested materials,
the composites should be recommended for similar
applications instead of the polymer;
for two composites (15% and 25% GF),
results pointed out specific processes characterizing
composites with short fibers: wear decreases when
average pressure increase as a result of compressing
the tribolayer, PTFE remaining kept in a non-
arranged fiber net. Also, small changes in the fiber
net allow capturing water drop or impurities, so
sample mass may increases (see Fig. 4a and [5]).
Tribological parameters as friction coefficient
and wear of composites with PTFE matrix depend
on tribotesters by geometrical shape and dimensions
[4,8], but processes within the superficial layers are
similar, fact proved by SEM images or 3D
profilometry analysis, even if for polymeric
composites the studies are still a few [1,4,6].
3.2 Analysis of several 3D parameters of surface
topography
Profilometer PRO500 3D (with stylus) was
used to measure the surface topography [16] assisted
by a dedicated soft [9]. The selection of area size is
important since this should be large enough to
characterize a representative part of the surface or at
least to generate stable parameter values. Here there
were chosen zones in the plate center. The vertical
range was set at 500 m and the scan speed was
selected as 35 m/s. All records have been done
with 200 points on each line. Pitch between lines
was set at 5 m. All 3D parameters were calculated
for raw profiles because they offer the possibility of
pointing out extreme values [10], this being one of
the aim of the paper: to detect extreme values of the
analyzed parameters and, as it is written in [1, 9] the
raw profiles help building a virtual image closer to
the actual one. The equivalent contact force of the
stylus was set for polymeric surfaces, at 16 mg.
There were analyzed here only some of 3D
amplitude parameters: the roughness average Sa
[m], the root mean square (RMS) parameter, Sq
[m], the surface skewness, Ssk [-], the surface
kurtosis, Sku [-], the peak-peak height [m] and
three parameters obtained based on the bearing area
curve: the reduced summit height, Spk [m], the
core roughness depth, Sk [m], the reduced valley
depth, Svk [m], as defined in [9]. The wear value
has a minimum for ~25% GF, but only Ssk has an
evolution that could be related to the wear one: Ssk
seems to be a mirror of wear evolution as it has a
maximum in the same range where the wear is
minimum. Sku plot has a similar shape as for wear,
but the point obtained for 40% glass fiber does not
confirm this tendency (Fig. 5).
v=2.5 m/s
p (MPa)
-0.02
0
0.02
0.04
0.06
0.08
0.1
0 10 20 30 40
Glass fiber concent rat ion (%)
W
e
a
r

(
g
)
0.22
0.77
1.26
2.02
a) Wear as function of glass concentration
v=2.5 m/s
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
0.05
0 10 20 30 40
Time (min)
F
r
i
c
t
i
o
n

c
o
e
f
f
i
c
i
e
n
t
PTFE
PTFE = 15% GF
PTFE+ 25% GF
PTFE + 40% GF
b) Friction coefficient
Figure 4. Tribological behavior of tested materials
116
Sampl e F1
0
0.2
0.4
0.6
0.8
1
1.2
Sa Sq
(
m
i
c
r
o
n
s
)
Sampl e F1
-5
0
5
10
15
20
25
30
Ssk Sku Sy
min (1-5)
Average F1-(2)
Average F1-(3)
Average F1-(5)
MAX (1-5)
Sampl e F1
0
0.5
1
1.5
2
Spk Sk Svk
(
m
i
c
r
o
n
s
)
Sampl e G1
0
0.2
0.4
0.6
0.8
1
1.2
Sa Sq
(
m
i
c
r
o
n
s
)
Sampl e G1
-5
0
5
10
15
20
25
30
Ssk Sku Sy
min (1-5)
Average G1-(2)
Average G1-(3)
Average G1-(5)
MAX (1-5)
Sampl e G1
0
0.5
1
1.5
2
Spk Sk Svk
(
m
i
c
r
o
n
s
)
Sampl e A1
0
0.2
0.4
0.6
0.8
1
1.2
Sa Sq
(
m
i
c
r
o
n
s
)
Sampl e A1
-10
0
10
20
30
40
50
60
70
80
90
Ssk Sku Sy
min (1-5)
Average A1-(2)
Average A1-(3)
Average A1-(5)
MAX (1-5)
Sampl e A1
0
0.5
1
1.5
2
Spk Sk Svk
(
m
i
c
r
o
n
s
)
Sampl e B1
0
0.2
0.4
0.6
0.8
1
1.2
Sa Sq
(
m
i
c
r
o
n
s
)
Sampl e B1
-5
0
5
10
15
20
25
30
Ssk Sku Sy
min (1-5)
Average B1-(2)
Average B1-(3)
Average B1-(5)
MAX (1-5)
Sampl e B1
0
0.5
1
1.5
2
Spk Sk Svk
(
m
i
c
r
o
n
s
)
Figure 5. 3D parameters for the studied plates: F PTFE, G PTFE + 15% GF, A PTFE + 25% GF, B
PTFE + 40% GF; GF glass fibers
As Ssk< 0, it may be a bearing surface with
holes and its high values may indicate extreme holes
or peaks on the surface. Sku being higher than 3,
reflects a surface with high centered distributions of
peaks. Average values and up and down deviations
for the 5 measurements on the studied samples are
given in Table 1.
By analyzing values of 3D parameters for the
tested materials and conditions it could be concluded
that surface is still smooth enough to continue the
tribosystem functioning, but there are insulated
micro-zones with higher maximum values, which
could reveal the fibers fracturing (see values for Sy,
Spk).
117
It could be concluded that for assessing the
quality of worn surfaces there is not possible to
apply rules and recommendations given in [10] and
each research should be adapted taking into account
the tribosystem, including materials in contact,
triboelements shapes, regime (dry, lubricated,
boundary lubricated), movement type, environment
requirements.
Table 1. Average values and up and down deviations for the 5 measurements
Material Range of deviations for studied parameters
PTFE
41.0% 31.9% 76.0% 90.4% 78.1%
88,6% 80.1% 133.1% 132.2% y 132.2%
32.7% 36.0% 73.0%
94.6% 83.4% 119.9%
Sa=0,32 ; Sq 0.41 ; Ssk 0.58 ; Sku 5.4 ; S 4.5 ;
Svk 0.49 ; Sk 0.95 ; Spk 0.35






PTFE + 15% glass fibres
62.3% 50.3% 49.9% 140.7% 44.0%
104,5% 91.4% 196.9% 203.5% y 90.1%
39.6% 88.0% 13.5%
79.6% 121.9% 30.1%
Sa=0,30 ; Sq 0.41 ; Ssk 1.43 ; Sku 9.46 ; S 4.65 ;
Svk 0.61 ; Sk 0.77 ; Spk 0.26






PTFE + 25% glass fibres
43.2% 61.2% 66.4% 159.4% 60.9%
78,8% 103.0% 170.5% 234.9% y 108.2%
91.0% 64.3% 66.7%
135.2% 98.2% 110.1%
Sa=0,42 ; Sq 0.63 ; Ssk 3.33 ; Sku 35.45 ; S 10.2 ;
Svk 0.97 ; Sk 1.17 ; Spk 0.45






PTFE + 40% glass fibres
8.5% 25.8% 31.5% 93.9% 46.6%
16 ,3% 40.6% 97.5% 133.7% y 68.5%
33.5% 9.6% 46.6%
54.8% 20.6% 73.6%
Sa=0,59 ; Sq 0.90 ; Ssk 2.50 ; Sku 16.80 ; S 11.77 ;
Svk 1.64 ; Sk 1.48 ; Spk 0.69






4. CONCLUSIONS
For tested sliding speeds, the wear tendency
is similar for the tested material, but values for mass
loss are almost one order less for the composites as
compared to PTFE. For higher speed (2.5 m/s)
composites wear are reduced due to a synergic
effect of at least a partial water film and the polymer
compression into the non-uniform fiber net. The
high wear gradient between 0% GF (the polymer)
and 610% GF suggests that at a lower speed, the
composites offer conditions for a partial EHD
regime. The presence of a fluid film is proved by
both the very reduced wear, even if using a poor
lubricant as water [4, 7] and the very low values of
the friction coefficient.
Wear has been strongly influenced by fiber
concentration (see Fig. 4a). Without fibers, the thin
micro-bands of polymer are detached with high rate,
the water, especially at higher pressure, promoting
tearing of the material, rolling and rapid movement
of this debris outside the contact. Even a rare and
random net of fiber keep the polymer of being
peeled, rolled and detached from the surface.
By analyzing the variation of average values
as a function of the measurement number (fig. 5) it
is obvious that 23 measurements are not
representative at least for the studied surfaces, but 5
measurements have given a good indication of the
surface quality, especially if this assessment is not
reduced to studying Sa parameter. Comparing the
average and extreme values for the analyzed
amplitude 3D parameters, one may notice two
distinct groups (see also Table 1):
the group of Sa, Sq and Sy that have a slight
tendency to increase when the GF concentration
increases, but with measured values spread in a not
so large range around the average value,
the group of Sku and Ssk that spread on a
large range.
By analyzing the functional parameters the
following conclusions could be drawn:
Svk is slightly increasing when the GF
concentration increases, but the value of 1.6 m for
the composite with 40% GF means that many glass
fibers could remain outside the matrix being sources
of micro-abrasions;
Sk is the functional parameter with higher
values as compared to other ones, Spk and Svk,
meaning a good bearing core zone of the superficial
layers for all tested materials, the lowest values
being obtained for the composite with 15% GF;
Spk has a large variation for the polymer,
logically because of tearing off the polymer and of
re-bonding of the polymeric debris, but when
adding glass fibers this parameters becomes lower,
especially for 15% and 25% GF;
the highest values for these parameters were
obtained for the composite with the highest glass
fiber concentration (40%wt), but wear of this
composite (see Figure 4a) still recommends it for
actual applications with water lubrication, high
speed and average pressure around 2 MPa.
These results underline the possibility of
relating 3D roughness parameters to the tribological
ones (wear, friction coefficient etc.) for polymeric
composites, too. But data should be enough
numerous in order to estimate with high degree of
confidence the surface quality with the help of 3D
roughness parameters.
118
Sampl e F1
-3
-2
-1
0
1
2
Sa Sq Ssk Svk Sk Spk
(
m
i
c
r
o
n
s
)
min
MAX
Average F1-(5)
Sampl e F1
0
5
10
15
20
25
30
35
Sku Sy
PTFE
Sampl e G1
-5
-4
-3
-2
-1
0
1
2
Sa Sq Ssk Svk Sk Spk
(
m
i
c
r
o
n
s
)
min
MAX
Average G1-(5)
Sampl e G1
0
5
10
15
20
25
30
35
Sku Sy
PTFE + 15% GF
Sampl e A1
-7
-6
-5
-4
-3
-2
-1
0
1
2
Sa Ssk Sk
(
m
i
c
r
o
n
s
)
min
MAX
Average F1-(5)
Sampl e A1
0
10
20
30
40
50
60
70
80
90
Sku Sy
PTFE + 25% GF
Sampl e B1
-5
-4
-3
-2
-1
0
1
2
Sa Sq Ssk Svk Sk Spk
(
m
i
c
r
o
n
s
)
min
MAX
Average B1-(5)
Sampl e B1
0
5
10
15
20
25
30
35
Sku Sy
PTFE + 40% GF
Figure 6. Amplitude and hybrid parameters for one plate from a three-plate set: average of all 5
measurements and the up and down deviations from this average value
119
REFERENCES
1. Blunt L., Jiang X., 2003, Advanced techniques
for assessment surface topography, Elsevier.
2. Bratcu O., Tomescu (Deleanu) L., Bologa O.,
2002, Tribological Behaviour of PTFE + Glass
Fibber Composites Used for Axial Bearings under
Water Lubrication, Analele Universitii Dunrea
de Jos din Galai, Fascicle VIII, Tribology, pp. 61-
65.
3. Burris L.D., Sawyer G.W., 2006, A Low
Friction and Ultra Low Wear Rate PEEK/PTFE
Composite, Wear, 261, pp. 410-418.
4. Dasari A., Zu Z.-Z., Mai Z.-W., 2009,
Fundamental Aspects and Recent Progress on
Wear/Scratch Damage in Polymer Nano-
Composites, Materials Science and Engineering R,
63, pp. 3180.
5. Deleanu L., Brsan I.G., Andrei A., Rp M.,
Diaconu N., 2008, PTFE Composites and Water
Lubrication. II. Surface Characterisation, Revista
de Materiale plastice, vol. 4, pp. 332-338.
6. Khedkar J., Negulescu I., Meletis E.I., 2002,
Sliding wear behavior of PTFE composites,
Wear, 252, pp. 361369.
7. Larsen T., Andersen T.L., Thorning B.,
Horsewell A., Vigild M.E., 2008, Changes in the
Tribological Behavior of an Epoxy Resin by
Incorporating CuO Nanoparticles and PTFE
Microparticles, Wear, vol. 265, n
o
1-2, pp. 203-
213.
8. Sawyer G.W., Freudenberg K.D., Bhimaraj
P., Schadler L.S., 2003, A Study on the Friction
and Wear Behavior of PTFE Filled with Alumina
Nanoparticles, Wear, 254, 573580
9. **** The Scanning Probe Image Processor
SPIP
TM
, Version 4.7 (2008).
10. **** SR SR EN ISO 4288:2002 Geometrical
product specifications (GPS) - Surface Texture:
Rules and Procedures for the Assessment of
Surface Texture.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 120-127
Minodora RP
e-mail: minodora.ripa@ugal.ro
Simona BOICIUC
University Dunarea de Jos of Galati,
ROMANIA
CHARACTERISATION OF LASER CLADDING
WITH NICrBFe Al ALLOY BY
PROFILOMETRIC STUDY OF THE SCRATCH
TRACKS
The sliding indentation test have good results for characterizing
metals and alloys, polymers, ceramics, composites and a great range
of coatings, and often it is connected with wear tests and modeling
and simulation techniques. This paper presents research results on
several multi-layer claddings achieved by powder injection of Ni
alloy from the Ni-Cr-B-Fe-Al system, in the bath melt by CO
2
laser
in continuous wave. The comparisons of the geometrical
characteristics of the different digital depth profiles confirm the
better scratch behavior of the laser cladding layers.
Keywords: laser cladding, sliding indentation, wear track
1. INTRODUCTION
Laser surface treatments have become a
profitable alternative to conventional surface
processing technologies in many applications, and
the laser has become a valuable and cost effective
tool. Laser surfacing offers a clean and reliable
method of depositing coatings onto substrates,
especially in order to increase wear and corrosion
resistance.
Laser cladding is a high precision technique
to generate desired surface properties, whilst
retaining the mechanical properties of the substrate
[1-4]. The most frequently used cladding materials
are the powders in the single step processed (blown
powder). The two step process has the advantage of
very low dilution, but its use is limited to almost flat
surfaces. The blown powder process is used more
by industry, due to its better flexibility with respect
to surface geometry. It is also easier to blend
powders for a required chemical composition.
In order to maintain the genuine properties of
the clad material, only a very thin layer of the
substrate must be melted to obtain the minimum
dilution (0,5 - 3%) of the metallurgical bond of the
additional material with the substrate. The structure
and the properties depend on the melting
temperatures of both the support and clad material,
their chemical composition and they may vary by
applying various thermal regimes and granulation of
the powder added [5,6].
This paper presents research results on
several multi-layer claddings achieved by powder
injection of Ni alloy from the Ni-Cr-B-Fe-Al
system, in the bath melted by CO
2
laser in
continuous wave.
Wear and friction of sliding components are
highly related to their resistances to contact
deformation and damage [7].
A scratch test combined with an instrumented
indentation test is a very useful tool in examining the
microscopic surface deformation mechanisms and
processes that are taking place under mechanical
contact/sliding. The scratch test was first suggested
for coating adhesion measurements more than thirty
years ago. The scratch testing method is today
widely used, especially by the coating industry and
coating development laboratories, as well as in
research for evaluating the tribological properties of
coatings and other hard surfaces [8]. Different
standard were elaborated in Europe and USA.
The scratch test gives good results for
characterizing metals and alloys, polymers,
ceramics, composites and a great range of coatings
[7-9,10] and often it is connected with wear tests [9,
11-13] and modeling and simulation techniques
[8,11,13].
2. LASER CLADDING EXPERIMENTAL
RESEARCHES
The powder used for laser cladding, Alliages
Speciaux 7569 Alliajes Frittes, has the following
chemical composition (wt.%): 8.9%Cr; 4.5%Fe;
5.1%B; 2.4%Al; 0.6% Cu; balance Ni [2,3]. Grain
fractions from 80-90 m range were separately
screened in order to be used as addition material.
Powder had a spherical shape, which provided a
fluid flow of addition material through the injection
system. Before the addition of the material feeding
121
into the system tank, powder was dried at 110
o
C
temperature for 15 minutes [3].
Cladding was performed on a 1C45, SR EN
10083-1:1994 steel specimen, by a Laser GT 1400W
(Romania) type CO
2
continuous wave equipment,
with x-y-z coordinate running table and computer
programmed running. This equipment, provided by
powder injection system on the laser melt surface,
was updated at UZINSIDER Engineering, Galai,
Romania.
After adjusting the power level of laser
radiation and laser beam diameter on the specimen
surface, claddings were carried out under the form of
parallel strips partly overlapping, with a transverse
advance step of 1,5 mm. Final layer thickness was
the result of overlapping 4-5 layers.
To determine the optimum laser cladding, the
flow rate of material added, the surface scanning
speed and the initial specimen temperature were
varied. Researches on different working regimes
(working conditions and the thickness of the clad
layers) were performed.
Table 1 shows the characteristics of the
optimal cladding regime, which provides the highest
hardness and thickness of the surface layer.
In order to characterize the exploitation
behavior of the clad layers, for applications
requiring wear and corrosion resistant surfaces [4-
6,13] the following tests were performed:
determination of the thermal stability; wear test on
rotary disk with abrasive paper (STAS 9639-81
Romanian Standard); corrosion tests; scratch test;
profilometric studies of the scratch tracks.
The specimens realized with this regime were
characterized as follows: macro and microstructural
analyses (fig. 1); hardness (HV
5
); microhardness
(HV
0,1
); phase quality analysis by X ray
difractometry (DRON 3 Difractometer); EDX
microanalysis of the clad Ni alloy composition
(SEM XL 30ESEM TMP - Phillips, spectrometer
EDS - EDAX Saphire).
Table 1. Working regime used in laser cladding
NOTE: P - laser radiation power , v scanning speed of the laser beam on the processed surface, d
s

diameter of the laser beam; p

av
- transversal advance step, g - thickness of clad layers; m
p
- flow rate
of added material.
a. b.
Figure 1. Microstructure of clad nickel based alloy (v = 7,5 mm /s, m
p
= 105 mg /s)
a) base of clad layer; b) surface layer . Electrolyte attack, solution 50% HNO
3
[1,3,14]
Working regime
Added
material rate
[mg/s]
No. of
overlapping
runs
P [W]
v
[mm/s]
d
s
[mm]
p
av
[mm]
g
[mm]
Hardness
HV
5
[MPa]
105 4 1150 7,5 1,8 1,5 2,07 11450
122
3. SCRATCH TEST
The scratching is a physical process during
which a sharp object is pressed onto, and drawn over
the surface of the coating simultaneously. The
normal load is either kept constant or progressively
increased, depending on the purpose of the test and
machine availability. During constant load scratch
testing, the normal force on the scratch indenter is
maintained at a constant level. Multiple tests at
increased constant load levels can be used to
determine the critical scratch load or scratch
hardness. The constant load scratch requires more
tests to find the critical point of damage, but it can
be carried out in a less costly test rig compared to
the progressive loading scratch test. It can also be
used to detect non-uniformity of the coating over the
entire surface. In a scratch test, as a result of friction, a
tangential load is added to the normal load. This
friction traction superposes a compressive stress at the
front edge of the contact and a tensile stress at the
trailing edge [15].
3.1. Scratch rig and methodology
The scheme of the rig for scratching tests
(sliding indentation) is presented in Figure 2. It has
the following components: 1-frame, 2-ABB
frequency converter, 3-electric engine, 4-elastic
coupling, 5, 6, 7, 8-mechanical transmission, 9-
horizontal column, 10-balls guiding, 11-vertical
column, 12-balls guiding, 13-force transducer, 14-
elastic system, 15-loading screw, 16- specimen, 17-
sustenance surface, 18- specimen fixing device, 19-
balls guiding, 20-force transducer, 21-screw for
transversal movement, 22- indenter.
Figure 2. The components of the rig for scratching tests (sliding indentation) [16,17]
Because of the components of the mechanical
transmission and the presence of the frequency
converter, the speed of the column 9 is very slow
and varies between 0 and 17 mm/min. In this
experiment, the speed is setted at 0.2 mm/s. Under
this speed condition, the deformations are quasi-
steady. The measuring system is composed by two
force transducers (050 kN) and a data acquisition
system. The indenter is a steel ball with a diameter
of 12.675 mm. The normal forces used for
indentation were: F1 = 2.886 kN; F2 = 4.330 kN; F3
= 5.773 kN; F4 = 7.216 kN.
3.2. Specimens
The substrate material used in this study is
1C45 steel grade with the following chemical
compositions (wt.%): 13.75Ni, 2.72Mo, 0.019C,
0.50Si, 1.87Mn, 0.012S, 0.022P, 17.43Cr,
0.002Ni, 0.002Al.
Hardness differences between the center and
edge of he laser clad layer were found. The center
has higher hardness and the margins slightly lower
hardness. This is due to more intense heating of the
edges, thus increasing the specimen surface
temperature and increased evaporation processes.
Three specimens was tested: 1. code MB
specimen made of substrate material; code A -
specimen from the center; B - specimen from the
margins.
The presence of the intermetalic components
(borides, carbides) cause the adherence inhibition.
Their amount increases from the substrate material
containing precipitates of cementite, for the
specimen A, possessing a larger amount of borides.
The maximum will be reached in the case of the
specimen B, which has the highest hardness. Thus,
the deformation behavior of the three specimens will
be different.
The specimen surfaces were manufactured by
grinding, perpendicular to the length. Experimental
observations confirmed that this leads to an
appropriate asperities orientation, in order to obtain
very accurate wear scars. This is due to the plastic
deformation (Figure 3) which could be measured
with greater precision.
123
Figure 3. Picture of specimen MB with the scratch tracks
4. PROFILOMETRIC STUDY OF THE
SCRATCH TRACKS
A stylus digital profilometer SURTRONIC
3+ (Taylor-Hobson) was used to assess the
topography changes caused by scratching on laser
cladding layer. For all the specimens, the average
surface roughness was about Ra 0.210m. Figure
4 shows relevant roughness profiles for each
specimen.
The specimen with scratching tracks is put on
a support and guiding element (fig. 5) and could
have a translation movement along its own axis.
The stylus of the profilometer has a transversal
movement over the track. As a result of this
guiding, the cross-sections are parallel.
-5
-3
-1
1
3
5
0
5
0
0
1
0
0
0
1
5
0
0
2
0
0
0
2
5
0
0
3
0
0
0
3
5
0
0
4
0
0
0

m
a) substrat material (code MB)
-5
-3
-1
1
3
5
0
5
0
0
1
0
0
0
1
5
0
0
2
0
0
0
2
5
0
0
3
0
0
0
3
5
0
0
4
0
0
0

m
b) cladded specimen (code A)
-5
-3
-1
1
3
5
0
5
0
0
1
0
0
0
1
5
0
0
2
0
0
0
2
5
0
0
3
0
0
0
3
5
0
0
4
0
0
0

m
b) cladded specimen (code B)
Figure 4. Roughness profile of specimens, before scratch test
Figure 5. The support and guiding element: 1 stylus of the profilometer,
2 - sliding indentation tracks, 3 - specimen, 4 - support and guiding element
124
-50
-40
-30
-20
-10
0
10
20
30
40
50
0
5
0
0
1
0
0
0
1
5
0
0
2
0
0
0
2
5
0
0
3
0
0
0
3
5
0
0
4
0
0
0
m
m
MB; F1
MB; F2
MB; F3
MB; F4
a)
-50
-40
-30
-20
-10
0
10
20
30
40
50
0
5
0
0
1
0
0
0
1
5
0
0
2
0
0
0
2
5
0
0
3
0
0
0
3
5
0
0
m
m
A; F1
A; F2
A; F3
A; F4
b)
-50
-40
-30
-20
-10
0
10
20
30
40
50
0
5
0
0
1
0
0
0
1
5
0
0
2
0
0
0
2
5
0
0
3
0
0
0
3
5
0
0
m
m
B; F1
B; F2
B; F3
B; F4
c)
Figure 6. Comparison of the wear track depth profiles, (for the same specimen four plots are compared).
Normal forces: F1=2.886 kN, F2=4.330 kN, F3=5.773 kN, F4=7.216 kN
-50
-40
-30
-20
-10
0
10
20
30
40
50
0
5
0
0
1
0
0
0
1
5
0
0
2
0
0
0
2
5
0
0
3
0
0
0
3
5
0
0
4
0
0
0
m
m
MB; F1
A; F1
B; F1
-50
-40
-30
-20
-10
0
10
20
30
40
50
0
5
0
0
1
0
0
0
1
5
0
0
2
0
0
0
2
5
0
0
3
0
0
0
3
5
0
0
4
0
0
0
m
m
MB; F2
A; F2
B; F2
-50
-40
-30
-20
-10
0
10
20
30
40
50
0
5
0
0
1
0
0
0
1
5
0
0
2
0
0
0
2
5
0
0
3
0
0
0
3
5
0
0
4
0
0
0
m
m
MB; F3
A; F3
B; F3 -50
-40
-30
-20
-10
0
10
20
30
40
50
0
5
0
0
1
0
0
0
1
5
0
0
2
0
0
0
2
5
0
0
3
0
0
0
3
5
0
0
4
0
0
0
m
m
MB; F4
A; F4
B; F4
Figure 7. Comparison of the wear track depth profiles for the same force.
Normal forces: F1=2.886 kN, F2=4.330 kN, F3=5.773 kN, F4=7.216 kN
125
The shape of the scratch tracks was obtained
by measuring six depth profiles across each track
with the stylus digital profilometer SURTRONIC
3+.
For the specimen code MB, the material is
advanced plasticly deformed, the indenter is in deep
penetration; as consequence the friction surface is
large and the adhesion tendency is high. Thus, for
the basic material the increasing of normal force
could lead to the increasing of the coefficient of
friction. Analyzing the specimen A, it can be seen
that plastic strain ratio is lower due to higher
amounts of precipitates comparing to the case of the
substrate material, resulting a lower penetration of
the indenter and a lower friction surface associated
with a lower adhesion trend.
In the case of the specimen B, it appears that
high hardness, due to the large amount of borides
make the plastic deformation to be minimal. The
indenter penetration into the material is the lowest
of the three cases.
Analyzing the depth profiles presented in fig.
6 some observations may be made:
for the substrate material MB, the increasing
of lateral ridges occurs with the increasing of normal
force.
due to the increasing of the normal force,
material A is deforming less under the action of the
ball, on both width and depth as comparing with
substrate material and lateral ridges are more
flattened.
the specimen material code B, which has a
higher hardness and yield limit than the specimen A,
shows a lower deformation on both depth and width,
lateral ridges being lower.
Depth profiles plotted in Figure 7 present a
comparison of the plastic deformations of the
investigated materials, on the same normal load.
Due to the increasing hardness layer, the profile
depth reduces, as well as the height of the lateral
ridges.
This fact indicates that the clad laser surface
has a higher resistance related to plastic deformation
than the substrate material.
Figure 8 presents the geometrical
characteristics of a depth profile, computed by the
software of the profilometer.
Table 2 shows the values of the width, the
maximum depth and the cross-sectional area of the
wear track for the specimen code A.
Figure 8. Geometrical characteristics of a depth profile
Figure 9. Track depth variation versus normal force
126
Table 2. Example: values of geometrical characteristics of the depth profile, specimen code A
Profile code
Normal
force
Width [mm]
Maximum
depth
[m]
Cross-sectional area of
the wear track [m
2
]
A1_u_1 1 10.1 5953
A1_u_2 1.01 10.7 6045
A1_u_3 1.01 9.91 6193
A1_u_4 1 11.1 7357
A1_u_5 1.11 9.88 6857
A1_u_6 0.86 6.72 3437
A1_u_7 0.93 9.16 5278
A1_u_8 0.94 7.28 3827
A1_u_9
F1
0.95 8.14 4087
A2_u_1 1.33 17.5 14342
A2_u_2 1.33 18.3 13849
A2_u_4 1.25 19.7 14963
A2_u_5 1.23 20.2 14158
A2_u_6 1.24 15.5 11266
A2_u_7 1.29 20.2 15605
A2_u_8 1.33 17.3 12532
A2_u_9 1.27 19.1 13180
A2_u_10
F2
1.28 14.8 11196
A3_u_1 1.5 23.5 22101
A3_u_2 1.46 24.2 22265
A3_u_3 1.47 23.3 21020
A3_u_4 1.51 24.6 22941
A3_u_5 1.46 22.8 21298
A3_u_6
F3
1.46 25.3 22218
A4_u_1 1.64 34 35704
A4_u_2 1.58 36 36262
A4_u_3 1.6 33.2 33981
A4_u_4 1.539 36.3 35830
A4_u_5 1.62 38.7 39123
A4_u_6
F4
1.64 39.5 39129
Figure 10. Track width variation versus normal force
127
Figure 9 presents track depth variation with
normal force and Figure 10 shows track width
variation with normal force. In Figure 9 it could be
notice that for small normal forces the deformation
depth of specimen code B is reduced but the zones
near to the track begin to participate at the
deformation process, recording a maximum width,
in good correspondence with fig. 6.
With the increasing of the normal force, in
depth deformation becomes prevalent and for the
force F4 the width for specimen B get less than the
width of the specimen A.
Analyzing the track width variation with the
normal force (fig. 10), it appears that trace depth
growth occurs due to normal force increasing and to
the arising of plastic deformation, a fact more visible
for the substrat material.
5. CONCLUSIONS
This paper presents the first step in the
complex characterization of laser cladding with Ni
CrBFe Al alloy by profilometric study of the
scratch tracks. The comparisons of the geometrical
characteristics of the different digital depth profiles
confirm the better scratch behavior of the laser
cladding layers. The researches will continue with
more complex tribological investigations, in order to
succeed a complete characterization of the properties
of the hard surfaces.
REFERENCES
1. Boiciuc, S. et al., 2009, EDX Analysis of Laser
Cladding Layers with Ni-Cr-B-Fe-Al Alloy,
Conference UGALMAT 2009, Galati, Romania.
2. Levcovici, D.T., Boiciuc, R., Levcovici, S.M.,
Gheorghie, C., 2006, Laser cladding of M2 Steel
on a steel substrate, The Intern, Thermal Spray
Conf. and Exposition (ITSC 2006) Seattle,
Washington, U.S.A, ASM Seattle 2006, Procs on
CD.
3. Levcovici, S.M., Levcovici, D.T., Gheorghie,
C., Boiciuc, S., 2006, Laser Cladding of Ni-Cr-B-
Fe-Al Alloy on a Steel Support, The International
Thermal Spray Conference and Exposition (ITSC
2006) May 15
th
17
th
, 2006, Seattle, Washington,
U.S.A, Proceedings on CD.
4. Liu, X.-B., Wang, H.-M., Microstructure and
Tribological Properties of Laser Clad /Cr
7
C
3
/TiC
Composite Coatings on -TiAl Intermetallic Alloy,
Wear 262 (5-6), pp. 514-521.
5. Gedda H., 2000, Laser Surface Cladding - A
Literature Survey, Lulea University of Technology,
Division of Materials Processing, Sweden.
6. Schneider, M.F., 1998, Laser Cladding with
Powder, Ph. D. Thesis, University of Twente,
Enschede, Holand/
7. Futami T., et al., 2009, Contact/Scratch-
Induced Surface Deformation and Damage of
CopperGraphite Particulate Composites, Carbon
47 (2009), pp. 2742 2751.
8. Holmberg K. et al., 2006, Tribological Contact
Analysis of a Rigid Ball Sliding on a Hard Coated
Surface. Part I: Modelling Stresses and Strains,
Surface & Coatings Technology 200, pp. 3793
3809.
9. Berns, H., Saltykova, A., 2009, Wear
resistance of in situ MMC produced by supersolidus
liquid phase sintering (SLPS), Wear, 267, pp.
17911797.
10. Wang, Z.Z., Gu, P., Zhang, Z., 2010,
Indentation and Scratch Behavior of Nano-
SiO2/Polycarbonate Composite Coating at the
Micro/Nano-Scale, Wear 269, pp.2125.
11. Avril L., 2003, Elaboration de revetements sur
acier inoxydable. Simulation de la fusion par
irradiation laser. Caracterisation structurale,
mecanique et tribologique, PhD Thesis, Ecole
Nationale Suprieure dArts et Mtiers, Centre
dAngers.
12. Emmerlich, J. et al., 2008, Micro and
Macroscale Tribological Behavior of Epitaxial
Ti3SiC2 Thin Films, Wear 264, pp. 914919.
13. Martukanitz, R.P., Babu S.S. and Vitek J.M.,
2004, Development of Advanced Wear and
Corrosion Resistant Systems through Laser Surface
Alloying and Materials Simulation, Applied
Research Laboratory, State College, PA 16804, Oak
Ridge National Laboratory.
14. Boiciuc, S., Levcovici, S., Levcovici, D.T.,
2007, Structural Modifications in Laser Cladding
Layers Heating up at Different Temperatures,
Metalurgia International, nr.8, 2007, pp.14-19, Ed.
Editura Stiintifica F.M.R.Bucharest, Romania.
15. Friedrich, k., Schlarb, A.K., 2008, Tribology
of Polymeric Nanocomposites. Friction and Wear of
Bulk Materials and Coatings, Tribology and
Interface Engineering Series, 55. Editor: B.J.
Briscoe, 2008.
16. Spnu C., 2002, Studii i cercetri pe
tribomodel privind deformatiile plastice in stratul
superficial la rostogolire si la alunecare, PhD
Thesis, University Dunrea de Jos of Galati (in
Romanian).
17. Spnu, C. et al., 2009, Sliding Indentation
Behaviour of The X 65 Hydrogenated Steel Grade,
The Annals of University Dunrea de Jos of
Galai, Fascicle VIII, Tribology, XV (2), pp. 109-
114.
ISSN 1220 - 8434 ACTA TRIBOLOGICA
Volume 18, (2010), 128-135
Monica VLASE
1
e-mail: monica_utcb@yahoo.com
Andrei TUDOR
2
e-mail: tudor@meca.omtr.pub.ro
1
Technical University of Civil Engineering of
Bucharest, ROMANIA
2
University POLITEHNICA of Bucharest,
ROMANIA
AN ANALYTICAL WEAR MODEL OF THE PIPES
FOR CONCRETE TRANSPORTATION
The flow of fresh concrete in the pipe can be realized only when the
concrete is saturated. The tribological solutions are formulated to
obtain the saturation of concrete. The effect of flow in pipe is
evaluated by the friction with the wall and the pipe wear. It is
defined a critical angle of concrete impact in transition between
horizontal and vertical pipe as a function to the friction coefficient,
the velocity and the mean radius of solid particle in fresh concrete.
The erosion wear model is proposed for inner wall pipe in contact
with concrete
Keywords: fresh concrete friction, erosion, wear model, fatigue
1. INTRODUCTION
Fresh concrete is a viscous two phases
mixture, in which the solid phase, (sand, gravel) is
suspended in the liquid phase, (cement binder and
water). The binder is the slurry (suspension medium)
of the solid phase and has a great importance in the
rheologycal behavior of fresh concrete [13].
In order to be pumped through the metallic
pipes, that is being pump able, fresh concrete must
fulfill the following conditions:
- All the aggregates must be surrounded by the
cement slurry which has also the role of binder, and
to move freely in this liquid phase. That means the
mixture must be saturated.
- To be able to form, near the solid wall of the
pipe, a lubricant layer of cement slurry and
aggregate with thin granulation.
- Under the concrete pumping pressure through
pipes, to avoid the appearance of the segregation
phenomenon, that is to avoid the separation of the
solid and liquid phases or the aggregate deposition.
The levels of saturated and unsaturated
concrete are suggestively presented in Figure 1. As
it is shown in Figure 1, segregated concrete is an
unsaturated mixture. The aim of this paper is to
define the condition of flow of the fresh concrete in
pipe and to analyze the effect of flowing on the wear
of wall.
2. UNSATURATED CONCRETE VELOCITY
Concrete flow through a circular, curved pipe
is different from a circular, straightlined one. In
order to analyze the flow average speed, it is
suggested a model of the concrete flow under the
action of weight forces and friction forces that
appear in the curved zone. Thus, it is considered an
horizontal, circular pipe with d
t
inner diameter, with
R
k
curvature radius under a |
e
angle, (Figure 2).
a)
b)
Figure 1. The saturation level of the concrete:
a) - saturated mixture; b) - segregated concrete
(unsaturated mixture), [3]
Regarding the concrete flow, the case of the
horizontal pipe with vertical curvature is the most
difficult one [4].
For a certain | angle, it is considered an
infinite small volume of concrete between the d|
elementary angle. The elementary concrete quantity
(dG) in this infinite small volume is:
g
k
m
Q
dG R d
v
= | , (1)
where Q
g
is the gravimetric flow rate (N/s) of the
pumped concrete, and v
m
is the average speed of the
concrete flow.
129
Figure 2. Concrete flow through quarted bond
(angle pipe)
The elementary concrete quantity (dG)
induces a certain pumping force in the flow
direction, corresponding to the | angle, and a
friction force (dF
2
) on the exterior or the interior
pipe surface, depending on the position of the
quarted bond (angle pipe) with the horizontal
direction
Regarding the circular displacement of the
concrete it will also appear centrifugal forces that
will determine friction in the contact zone:
2
m
fc e
k
dG
dF
g R
=
v
, (2)
where: g is the gravitational acceleration;
e
is the
friction coefficient of the concrete plug with the
pipe wall, at the exterior of the curvature.
Considering the mechanical equilibrium
conditions, it can be deduced the average speed of
the axial displacement of the concrete plug (v
m
)
and also, the angular speed of rotation of the
concrete plug (e
m
) into the pipe, under the form of
two differential equations of the fist order:
, )
m
e m k i m
1
d
g R sin cos 0
d

+ + | + | =
|
v
v v ; (3)
m k
e m i
m
d 6g R
6 cos 0
d
e
+ | =
|
v
v
. (4)
The solution of the differential equation (3)
(Bernoulli equation type, reducible to a linear
equation) is under the form:
2
m
am i e
2 2
mi rmi e
2
i e i e
e
e
1
e [(2 ) sin
(1 4 )
2 cos (1 2 )e cos ]
|
|

= = + | +
+
| + |
v
v
v v
(5)
where: v
mi
is the average speed of the concrete
plug displacement at the entrance of the curvature
zone (quarted bond), v
rmi
is the average relative
speed of the concrete plug at the entrance,
comparing with the average speed in gravitational
field of the concrete plug from a height equal with
the quarted bond height,
rmi mi k
2g R = v v .
The solution of the differential equation (4)
(linear, first order differential equation) is:
m k i
am
i t rmi m mi
0
e k m
t mi
0
6R cos
d
d ( )
12 R
d .
d
|
|
e |
e = = |
e
| |
|
|
\ .
}
}
v v v
v
v
(6)
By replacing the expression of the average
speed (v
m
/v
mi
) from (5) into (6) it is obtained the
angular rotation speed of the concrete plug in the
quarted bond zone, in which e
i
= v
mi
/R
k
is the
angular speed at the entrance of the curvature zone.
In Figure 3 it is presented the average speed
change of the concrete plug for a horizontal up to
a vertical curvature, upright (upwards).
Figure 3. Nondimensional average speed of the
concrete plug
The relative rotation of the concrete
concrete plug during the displacement into the
quarted bond is presented in Figure 4.
By analyzing the diagrams regarding the
average speed decrease in the curvature zone, it is
concluded the possibility of pipe blockage, meaning
that the concrete can not be pumped any more.
Thus, the necessary condition for a pumpable
concrete is that for a saturated concrete.
130
Figure 4. The relative rotation of the concrete
concrete plug in the pipes quarter bond
3. TRIBOLOGICAL CONDITIONS FOR
SATURATED CONCRETE
For a correct displacement of the concrete
into the pipe it is necessary to respect the continuity
and, implicit, a constant flow in any section. Thus,
the concrete, as a liquid phase in a solid phase
mixture, must flow in pipes with constant inner
diameter with constant average speed [4].
In order to establish the necessary pressure
gradient along the flow direction of the concrete (as
a whole), it is analyzed the case of a quarter bond
that bonds a horizontal pipe and an angled pipe in
vertical direction by |
e
angle (Figure 5).
The flow condition with constant flow rate
implies the force equilibrium along the flow
direction with constant speed (v
m
). Thus, it is
determined the pressure difference necessary for the
concrete flow in a quarted bond of a horizontal pipe
bond with a vertical angled pipe under an |
e
angle:
, )
i e 2 e i e
3 e e
p p p k cos sin 1
+2k .
A = = | | +
|
(7)
where
2 m k
k g R = , (
m
concrete density, g
gravity acceleration, R
k
average radius of the
quarted bond curvature), and
2
3 m m
k 2 = v is the
dynamic pressure in the pipe.
If we have the dimensionless pressure
difference Ap, against the dynamic pressure k
3
, it
results:
, )
i e k
a e i e
2
3 m
e e
p p 2g R
p cos sin 1
k
+2 .

A = = | | +
|
v (8)
Figure 6 presents the dimensionless pressure
difference as a function of the angle made by the
quarted bond in vertical plane with the horizontal
line.
Figure 5. The diagram of the forces for concrete
uniform flow in the quarter bond zone
Figure 6. The variation of the dimensionless
pressure drop in the pipes quarter bond
4. THE WEAR MODEL OF PIPE WITH
FRESH CONCRETE IN PUMPING PROCESS
The theory of quasistatic indentation can be
used for solid particle impact, which is in fresh
concrete. The impact speeds are much smaller than
the velocity of elastic and plastic deformation of
metallic materials.
On impact the deceleration of solid particle
generates the indentation force on the substrate. The
impact angle of solid particles in the pipes quarted
bond is variable. The dimensionless erosion rate (I
er
)
is defined as mass of material removed of pipe per
mass of eroding (solid particles in fresh concrete).
We accept the equation of motion of single
abrasive particle interacting with the surface
(Finnies models) [5]. The erosion of ductile or
brittle metals comprise two wear mechanisms
occurring simultaneously: one caused by cutting
action of free moving particles in fluid with impact
131
angle grater than the critical impact angle; other
caused by repeated elastic or plastic deformation
during collision with friction (MansonMiners rule)
[6].
Figure 7 shows the impact of fresh concrete
rigid particle with the pipe wall [7].
Figure 7. The impact of fresh concrete rigid particle
with wall of pipe (a), and the deformed volume (b)
The critical impact angle (|
cr
) is defined as
the angle of particle, which appears a microchip for
only one impact. This angle can be calculated by the
motion equation of particle and the mechanical
properties of target materials [6,7]:
- for the elastic contact between the rigid
particle of fresh concrete and the wall of pipe:
, )
5
2 2
2
c
0
0
, arcsin
cr
4 5
ab
(
o | | t u
(
| =
|
(

\ .
(

v
v
, (9)
where is the friction coefficient inside the pipe; v
0
the velocity of the solid particle into the fresh
concrete, as a function of the fluid velocity; oc the
yield strength of the pipe material; u the elasticity
parameter of pipe material;
ab
the density of solid
particle into the fresh concrete.
- for the plastic contact between the rigid
particle of the fresh concrete and the wall of pipe:
, )
2
0 c c
0
0 c
3 e 2 HB
, arcsin
crp
4 2 HB
ab
| |
o o
|
| =
| o +
\ .
v
v
,
(10)
where e
0
is the yield specific deformation of the pipe
material; HB the Brinell hardness of the pipe
material.
Figures 8 and 9 show the critical angle, as a
function of the impact velocity and the friction
coefficient.
When the impact angle of particle is smaller
than the critical angle, the dimensionless erosion
wear rate can be evaluated for threelimit positions
of the collision particles:
1)
0
tan( ) 1 | > ;
2)
0
0.5 tan( ) 1 s | s ;
3)
0
tan( ) 0.5 | s .
Figure 8. Critical angle vs. impact velocity of fresh concrete
132
Figure 9. Critical angle vs. friction coefficient of fresh concrete in pipe
The dimensionless erosion rate expression is
as follows [8]. For the cases of fresh concrete flow,
the erosion rate has the following equations:
- the elastic contact:
, )
, )
, )
t
m
ere 0 0
ab c
t 5
5
2
2
ab 0 0
e 0 0
8 4
I , , r, , t
3
5
sin
4
H , , r, , t ,
+
| |
| =
|
t o u
\ .
t | |
u |
|
\ .
|
v
v
v
(11)
where t is the fatigue parameter of the pipe material,

m
the density of pipe material and H
e
the
integral function which has three forms for the limit
condition of the collision of the fresh concrete
particles;
- the plastic contact
, )
, )
, )
t
c
c m
erp 0 0
ab
t 5
5
1 2
0 0 ab
c
p 0 0
2 HB
2 HB
8
I , , r, , t
3 0.5
1
2 sin
3
H , , r, , t .
+
| |
o +
|
o
|
| =
|

|
|
\ .
| |
| |
|
o
\ .
|
v
v
v
(12)
The integral functions H
e
or H
p
can be
evaluated by numerical methods. A comparison
between dimensionless erosion rate for all impact
angles of the solid spherical fresh concrete particle
in transition to horizontal to vertical pipe is given in
Figures 10 and 11 in elastic and plastic deformation.
Figure 10. Pipe erosion rate in elastic contact of solid fresh concrete particle vs. impact angle
133
Figure 11. Pipe erosion rate in plastic contact of solid particle fresh concrete vs. impact angle
Figure 12. Pipe erosion rate in elastic contact of solid particle fresh concrete vs. velocity
Figure 13. Pipe erosion rate in plastic contact of solid particle fresh concrete vs. velocity
134
Figure 14. Pipe erosion rate in elastic contact of solid particle fresh concrete vs. friction coefficient
Figure 15. Pipe erosion rate in plastic contact of solid particle fresh concrete vs. friction coefficient
The maximum erosion rate is a function of
impact angle of fresh concrete with the vertical
direction of pipe, function of friction with the wall
and function of velocity of fresh concrete in pipe.
The Figures 1215 show the effect of
velocity and friction coefficient on the erosion rate
in elastic and plastic deformation of pipes.
5. CONCLUSIONS
The concrete flow through a circular, curved
pipe is different from a circular, straightlined one.
Thus, the fresh concrete has an axial displacement
and angular speed and moves as a plug.
The dimensionless contact pressure of fresh
concrete increases drastically with the angle of the
horizontal pipe in the vertical direction.
The solid particle of fresh concrete acts
abrasive and deforms the pipe material. This effect
can be used to predict the erosion wear rate in elastic
or plastic deformation.
The position of maximum erosive wear rate
in the curved pipe is a function of friction and of
fresh concrete velocity.
135
REFERENCES
1. Balayssac, J.P., Detriche, CH., Grandet, J.,
1993 Interet de lessai dabsorbtion deau pour la
caracterisation du beton denrobage, Materials and
structures, 26.
2. Tattersall, G.H., 1983, The Rheology of Fresh
Concrete, Pitman Publish INC, 1983.
3. Thomas, N.L. ,Double, D.D., 1981, Calcium
and Silicon Concentrations in Solution During the
Early Hydration of Portland Cement and Tricalcium
Si, Cement and Concrete Research, 11, pp. 675
687.
4. Vlase, M., Tudor, A., 2009, Pumping and
Transport of Concrete Through Pipes (in
Romanian), Ed. BREN, Bucureti.
5. Finnie, I., McFadden, D.H., 1978, On the
Velocity Dependence of the Erosion of Ductile
Metals by Solid Particles at Low Angles of
Incidence, Wear 48, pp.181190.
6. Kraghelskii, I.V., Dobicin, M.N., Kombalov,
V.S., 1977, Osnovi rascetov na trenie i iznos,
Moskva, Mainostroenie.
7. Tudor, A., 2002, Frecarea i uzarea
materialelor, Editura Bren, Bucuresti, 2002.
8. Tudor, A., 2003, An ErosionCorrosion Wear
Model for the Ball Valve Crude Petroleum
Extraction Pump, Proceeding Nat. Trib. Conf.,
ROTRIB 03, 2426 sept, Galati.
136
(continued from outside back cover)
65 S. LE FLOCH, M.C. CORNECI, A.-M. TRUNFIO-SFARGHIU, M.-H.
MEURISSE, J.-P. RIEU, J. DUHAMEL, C. DAYOT, F. DANG, M. BOUVIER,
C. GODEAU, A. SAULOT, Y. BERTHIER
Imagerie Medicale pour Evaluer les Conditions du Fonctionnement
Tribologiques des Articulations Synoviales
77 M.C. CORNECI, A.-M. TRUNFIO-SFARGHIU, F. DEKKICHE,
Y. BERTHIER, M.-H. MEURISSE, J.-P. RIEU, M. LAGARDE,
M. GUICHARDANT
Phospholipides dans le Fluid Synovial - Influence sur le Fonctionnement
Tribologique des Articulations Synoviales Pathologiques
85 I.C. ROMANU, E. DIACONESCU
Bioarticular Friction
89 A.-M. TRUNFIO-SFARGHIU, M.C. CORNECI, Y. BERTHIER,
M.-H. MEURISSE, J.-P. RIEU
Mechanical and Physicochemical Analysis of the Tribological Operation of
Joint Replacements
106 D. N. OLARU, C. STAMATE, A. DUMITRASCU, G. PRISACARU
Rolling Friction Torque in Microsystems
113 L. DELEANU, S. CIORTAN
Evaluating Tribological Damages by 3D Profilometry
120 M. RP, S. BOICIUC
Characterisation of Laser Cladding with NiCrBFeAl Alloy by
Profilometric Study of the Scratch Tracks
128 M. VLASE, A. TUDOR
An Analytical Wear Model of the Pipes for Concrete Transportation
ACTA TRIBOLOGICA VOLUME 18, 2010
CONTENTS
1 A. URZIC, S. CRETU
A Numerical Procedure to Generate Non-Gaussian Rough Surfaces
7 C. CIORNEI, E. DIACONESCU
Preliminary Theoretical Solution for Electric Contact Resistance between
Rough Surfaces
12 C.-I. BARBINTA, S. CRETU
The Influence of the Rail Inclination and Lateral Shift on Pressure Distribution
in Wheel - Rail Contact
19 C. SUCIU, E. DIACONESCU
Preliminary Theoretical Results upon Contact Pressure Assessment by Aid of
Reflectivity
27 S. SPINU
Numerical Simulation of Elastic-Plastic Contact
34 Y. NAGATA, R. GLOVNEA
Dielectric Properties of Grease Lubricants
42 J. PADGURSKAS, R. KREIVAITIS, A. KUPINSKAS, R. RUKUIA,
V. JANKAUSKAS, I. PROSYEVAS
Influence of Nanoparticles on Lubricity of Base Mineral Oil
46 A.V. RADULESCU, I. RADULESCU
Influence of the Rheometer Geometry on the Rheological Properties of
Industrial Lubricants
52 V.-F. ZEGREAN, E. DIACONESCU
Measurement of Lubricant Oil Microviscosity Based on Resonant Frequency
Shift of AFM Cantilever
58 M.C. CORNECI, A.-M. TRUNFIO-SFARGHIU, F. DEKKICHE, Y.
BERTHIER, M.-H. MEURISSE, J.-P. RIEU
Influence of Lubricant Physicochemical Properties on the Tribological
Operation of Fluid Phase Phospholipid Biomimetic Surfaces
(continued on page 136)