Escolar Documentos
Profissional Documentos
Cultura Documentos
www.elsevier.com/locate/surfcoat
Abstract
The tribological properties in surface contact can be markedly improved by coating the surface with soft or hard layers according to practical
applications. Rapid developments in surface engineering technologies have made available many surface engineering systems to combat diverse
component degradation problems and to meet the ever increasing demands for combined properties in modern machinery operating under ever
more severe conditions. Designers thus have many surface engineering technologies to select from in the design of components. Therefore,
numerical modelling is urgently required for surface engineering to assist engineers in their design of components with the surface and substrate as
a system. The present paper attempts to introduce a numerical approach for the contact behaviour of multi-layer systems and establish an initial
link between the contact stresses and performance of the surface engineered gears. Experimental tests have been carried out on untreated, TiN
coated, plasma nitrided and duplex treated steel gears. Here duplex treatment indicates plasma nitrided steel gear with TiN coating. The model has
also been applied in powertrain transmission titanium gears as well as polymer composite gear failure mechanisms.
© 2006 Elsevier B.V. All rights reserved.
zero, then a tangential traction is also present on the surface, in 2. Contact modelling
addition to the normal loading. This tangential traction induces
subsurface stresses in addition to those due to the normal load. This model is at present based on the following major as-
From their theory, it can be shown that the maximum shear sumptions, some of which may be relaxed as the work proceeds:
stress is moved towards the surface with the introduction of
friction (compared to Hertzians). This movement will be in- 1) The structure of the contact bodies is considered to be
creased as the coefficient of friction is increased and when the composed of one or more elastically homogeneous surface
coefficient of friction reaches some value the maximum shear layers rigidly bonded to each other and, in turn, to the
stress will reach the surface. substrate. Each component of the system, i.e. each layer and
The problem of indentation of a layered half-space consist- substrate, is taken as having its own distinct mechanical
ing of an elastic substrate with a thin elastic layer was first properties.
solved by Gupta and Walowit's model [5]. They derived the 2) The system is taken as being in contact with another elastic
mathematical formulation to the problem of a layered elastic body, which may be an elastic half-space or another multi-
surface indented by an elastic indenter and a rigid indenter, and layered body.
by using the Sneddon's integral transform techniques, numer- 3) The contact is assumed to give rise to a state of plane strain
ical solutions of the actual pressure distribution and the contact so that the system is considered in two dimensions only. This
zone were obtained. When the layer has a lower elastic modulus assumption holds well for line contacts, such as those which
than the substrate it is found that the actual contact pressure arise in gears and roller bearings.
distribution is very closely determined by a weighted sum of 4) Strains are assumed small, leading to the usual linear elastic
elliptic and parabolic functions. For a layer stiffer than the theory assumptions.
substrate the pressure distribution substantially deviates from an 5) The contact is considered to be dry, i.e., the presence of a
elliptical or a parabolic behaviour when the layer thickness is lubricant is neglected.
smaller than the half-contact width. However, when the stiffer
layer is thicker than the half-contact width, the pressure Fig. 1 shows an elastic half-space coated with n elastic
distribution closely follows the elliptic and parabolic functions. layers. Green's function for a unit normal load and a unit
With the advent of numerical techniques, the solution of tangential load for the generalised plane strain problem are
contacts no longer requires the geometric description of the derived. The layers are completely bonded to each other and to
body surfaces to be analytically convenient. In a more recent the elastic half-space they cover. A common method used in the
paper, Webster and Sayles [6] have provided a numerical model numerical solution of contact problems is to divide the contact
for the elastic contact of two-dimensional real rough surfaces. boundary into elements, over which it is assumed that the
The model used data directly recorded from a stylus measuring pressure is uniform [6]. Since the authors are using linear elastic
instrument. The significant effect of surface topography on the theory, each term of normal or horizontal movement is
pressure distribution can be seen. Depending on the topographic proportional to the element load, hence the authors can express
parameters, the maximum pressure in rough contact can be the displacement as the displacement due to unit uniform
many times higher than that in smooth contact, and the pressure elemental pressure times the real elemental pressure, resulting in
peaks arise from the highest asperities. Cole and Sayles [7] have the following expressions:
studied one layer coated real rough surfaces. In their work, a
numerical model which allows simulation of the two-dimen- ½U ¼ ½C½Pu ð1Þ
sional dry, frictionless, elastic contact of real rough layered
surfaces, without recourse to asperity models or topographical ½V ¼ ½D½Pv ð2Þ
statistics of the surface, is described. The model is shown to
reproduce smooth case behaviour to a high degree of accuracy. where
Further work has been done by Bell, Sun and Cole [8] for multi-
layered rough surface contact in real engineering surface con- [U] is normal displacement matrix
tacts. The computation of stresses in layered surfaces is of great [C] is influence coefficient for normal displacement
analytical and practical importance in many engineering appli- [Pu] is normal load matrix
cations where the friction and wear characteristics can be [V] is tangential displacement matrix
significantly improved by depositing thin layers of hard wear [D] is influence coefficient for tangential displacement
resistant materials, such as nitrides and carbides, on to the [Pv] is tangential load matrix.
surfaces.
In the past decade, significant progress has been made in the For a detailed description of the solution procedure of the
study of contact behaviour of layered surfaces. The underlying contact model, one may refer to the work of Mao, Sun and Bell
aims of the current research have been to understand the basic [9]. In summary, the surfaces are represented as arrays of points
mechanisms involved in various tribological processes of with a fixed distance, i.e., the sampling interval, a, apart, where
coated surfaces, to help in the selection of the optimum each point has associated with it a height value relative to some
coating/substrate combination system for a specific application datum. A master strain influence coefficient array is formed,
and in the design of engineering components with coatings. which holds the set of all profile point deflections that can occur
5798 K. Mao et al. / Surface & Coatings Technology 201 (2007) 5796–5803
Z l
1−m2i d 3 Gi 2−mi 2 dGi −jxyi dx
ux i ¼ x 3 − x e ð5dÞ
2kEi −l dxi 1−mi dxi x2
Z l
1−m2i d 2 Gi mi dx
uy i ¼ x 2 þ x Gi je−jxyi
2
ð5eÞ
2kEi −l dxi 1−mi x
where
where
In summary, every component in the system has its own Gi, 3.1. TiN/steel and duplex systems
and each Gi is described by four variables, the stress function
coefficients Ai, Bi, Ci, Di. Thus, for the n-layer system there are To demonstrate the capacity of the model to deal with multi-
4 (n + 1) variables. The condition of zero stress at an infinite layer systems under real rough and friction contacts, the model
depth in the substrate ensures that Cn+1 = Dn+1 = 0, and this has been used to simulate the contact of (a) a half-space (i.e.
reduces the number of variables to 4n + 2. The consideration of steel) coated with a 3 μm TiN coating and (b) a half-space (i.e.
the two types of loading at the surface yields two equations and steel) with duplex treatment (3 μm TiN and 8 μm Fe4N
the four stress and deflection equivalence considerations at coatings). Material properties used in the calculation were the
every interface in the system give a further set of 4n equations, following, elastic modulus: TiN, 640 GPa; Fe4N, 140 GPa;
thus a set of 4n + 2 simultaneous equations is obtained. The substrate, 210 GPa, Poisson's ratio: TiN, 0.2; Fe4N, 0.25;
equations are in terms of the elastic properties of every substrate, 0.3 [12]. Fig. 3 shows the rough surface contact
component in the system, the thickness of every layer in the pressure and the resultant subsurface stress (Von Mises stress)
system, the Fourier transforms of the applied loads and the distribution in the investigated TiN coating system. It has been
Fourier transform integrating variables. found that the effect of friction on the pressure distribution for
layered surfaces is very small [9] and thus Fig. 3(a) is applied to
3. Surface coatings evaluations all cases with different friction coefficients. However, the effect
of friction on the subsurface stress distribution is significant,
A computer code written in Fortran has been developed to particularly for rough surface contact (Fig. 3(b)). For smooth
simulate the specific contact problem based on the developed and frictionless contact, the maximum stress position is at
model. The digitised surface profile shown in Fig. 2 was certain distance below the surface. When a real rough surface is
obtained and taken as the surface of a layer. The technique considered, the subsurface stress distribution changes signifi-
developed by Webster and Sayles [6] was employed to deal with cantly: the maximum stress position is moved to the substrate
real rough surfaces. Computation results demonstrated that the surface even without friction. This is true for most of
surface pressure distribution for the real surface profile is engineering surface finish. When friction is introduced, the
significantly different from the traditional Hertzian smooth stress distributions change considerably: the maximum stress
contact theory, sometimes the pressure values in the case of real and the stress values in the near surface region increase with
surface profile are even twice as high as those of the smooth increasing friction coefficient. Clearly, in the case of hard
situation. coating systems, the combination of a rough surface and friction
Having found the surface deflections and pressures, the main force will greatly increase the level of subsurface stress and
area of interest in analysing the contact is usually the calculation move the maximum value to the surface, thus increasing the
of the stress states in the bodies induced by the contact. Detailed scope for failure.
derivation of various stress functions is given in Refs. [10,11]. The severe stressing conditions due to a single hard coating
In summary, a mesh is created in the body of interest (the layers of high elastic modulus can be improved by the duplex
and the substrate) and the stress at each point is calculated as the technique, involving plasma nitriding and PVD TiN coating.
sum of the stresses due to each pressure element on the surface. The iron nitride layer (Fe4N) formed during nitriding has a low
The input to the subsurface stress calculation programmes is elastic modulus and helps to reduce the stress level in the duplex
simply the output file from the contact simulation programme. system. Fig. 4 shows the rough surface contact pressure
The required information is layer thickness, material properties distribution and the resultant subsurface stress distribution for
and surface pressure distribution. The output is a file containing the duplex system. It can be seen that the distribution of pressure
5800 K. Mao et al. / Surface & Coatings Technology 201 (2007) 5796–5803
Tests were carried out using a speed ratio of 1:1 and the
tested gears run under dry condition as shown in Fig. 6. The
gears were run using a dedicated rig modified from a unit built
Fig. 3. TiN/steel system contact pressure (a) and resultant subsurface stress (b).
Table 1
Test results for gears with different treatments
Treatments Roughness change (Ra, μm) Weight loss (mg)
Untreated 7.026 550
TiN coated 5.539 250
Plasma nitrided 0.359 50
Duplex treated 0.035 10
Fig. 6. Gear test rig. Fig. 7. Schematic view of hardness/stress variations for different systems.
5802 K. Mao et al. / Surface & Coatings Technology 201 (2007) 5796–5803
Fig. 9. Strength/stress profiles for (a) TiN/Ti6Al4V and (b) TO/Ti6Al4V. Fig. 10. Tested titanium alloy gears.
K. Mao et al. / Surface & Coatings Technology 201 (2007) 5796–5803 5803