Professional Documents
Culture Documents
A R T I C L E I N F O A B S T R A C T
Article history: The influence of rectangle dimples with flat bottom on the friction of parallel surfaces at different sliding
Received 14 August 2009 conditions is investigated based on lubrication equations. The elastic deformation of rough surfaces is
Accepted 15 November 2009 evaluated using continuous convolution fast Fourier transform (CC-FFT). The friction coefficients for
Available online 23 December 2009
dimpled and non-dimpled parallel surfaces by simulation are compared with experimental results.
Results show that this kind of dimples can reduce the friction coefficient for cases with the smaller ratio
Keywords: of film thickness to roughness (h/Rq), small roughness or large applied load. The friction force for the
Friction coefficient
parallel surfaces can decrease due to the dimple effect over the range of the larger sliding speed, larger
Friction force
load or smoother surfaces.
Dimple
Parallel surfaces ß 2009 Elsevier B.V. All rights reserved.
Lubrication
CC-FFT
0169-4332/$ – see front matter ß 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.apsusc.2009.11.041
2864 F. Meng et al. / Applied Surface Science 256 (2010) 2863–2875
2. Experiments
Fig. 1. Schematic of the experimental setup for dimple friction test, where normal
The friction coefficient is first obtained through a parallel force is applied on the top of a steel cylinder, and the strip is pulled at a constant
surface friction testing. speed and pulling force is recorded.
F. Meng et al. / Applied Surface Science 256 (2010) 2863–2875 2865
8
< s 2 ð1 p Þ þ ð1 q Þ; ð p lx ; q ly Þ
Rðlx ; ly Þ ¼ lx ly (1)
:
0; ð p > lx ; q > ly Þ
Fig. 3. Laser textured sample surface; the cylinder diameter is 18.75 mm. Laser textured dimples with a width of 100 mm, length of 400 mm, and depth of 25 mm, and the
vertical and horizontal spaces are 460 mm and 770 mm between adjacent dimples.
2866 F. Meng et al. / Applied Surface Science 256 (2010) 2863–2875
Fig. 4. The friction coefficient by experiment with and without dimples at various sliding velocity at different normal forces F applied on the top of cylinder: (a) F = 60 N; (b)
F = 120 N; (c) F = 200 N.
following elastic deformation. In fact, FFT technique has been used surface characterization, respectively, as those of the generated
in solving the elastic deformation of rough surfaces, which can Gaussian surfaces. Meanwhile, the cylinder is assumed to be
shorten the computation time of deformation. Thus, using Eqs. (1)– stationary, while the strip slides with the sliding speed of U. The
(3) can yield Gaussian surfaces with various Perlenik number g. nominal film thicknesses between the cylinder bottom and strip is
Fig. 6 gives two generated surfaces respectively with g = 1/100, and denoted by h. If the roughness amplitude of the cylinder bottom
g = 100, which are separately comparable approximately to a and strip are represented by, respectively, the random variables d2
transverse and longitudinal surfaces. It can be seen from Fig. 6 that and d1, which are separately distributed with standard deviations
the orientations of the generated surfaces can be distinguished, Rq2 and Rq1 from their mean levels, the local film thickness hT can
which implies that the present code generating rough surfaces is be expressed as
satisfactory.
hT ¼ h þ d1 þ d2 þ V (4)
3.2. Basic equations where V is the composite elastic deformation of the cylinder
bottom and strip, whose expression will be further given.
Based on the above randomly generated Gaussian surfaces, With the above defined local film thickness hT, the pressure to
dimple effects on the friction of parallel surfaces can be simulated. be used for friction analyses can be solved through a Reynolds
In doing so, the cylinder bottom is assumed to be with the same equation. On the classical lubrication theory that the lubricant is
continuous, homogeneous and Newtonian fluid under isothermal
conditions, the two-dimensional Reynolds lubrication equation,
neglecting the dependency of lubricant density on the pressure and
the dependency of lubricant viscosity on the temperature, can be
described in the following form
@ h3T @P @ h3T @P U @hT @hT
þ ¼ þ (5)
@x 12m @x @y 12m @y 2 @x @t
W ¼ Wh þ Wa (10)
Fig. 7. Comparisons of the friction coefficients between transverse or longitudinal surfaces with and without dimples at various sliding speed when roughness deviation
Rq = 0.289 mm.
F. Meng et al. / Applied Surface Science 256 (2010) 2863–2875 2869
Same grid system as that used in generating a random rough Solving Eq. (15) for row by row solution is done with the under
surface in previous section is employed to solve the governing relaxation method. Meanwhile, a backward difference method of
Eq. (5). one order is used to achieve stable solutions. The pressure iteration
The coordinate original point is located at the beginning of precision is taken to be 0.0001. After each solution is obtained, the
cylinder bottom, i.e., inlet of lubricant. Meanwhile, to quickly solve relaxation coefficient a is used to update the values of pressure P in
Eq. (5) for the pressure, time-consuming calculation of pressure that row in the following form
variation with time is circumvented by the scheme that the
deviation of film thickness with respect to time on the right hand of P ¼ aP new þ ð1 aÞP old (16)
Eq. (5) is transformed as the deviation of film thickness with
respect to the space interval between two closely neighboring grid where Pold and Pnew are the pressure solutions before and after
points [13]. The details of transmission will not be repeated here. every iteration, respectively.
Thus, the discredited equation of Eq. (5) in difference form can be When the applied load F is balanced by the load-carrying
expressed as capacity W, the computation terminates. During the process, CC-
FFT technology is used to evaluate the elastic deformation
Ei; j P i; j þ Ai; j Pi1; j þ Bi; j Piþ1; j þ C i; j Pi; j1 þ Di; j P i; jþ1 F i; j P i; j ¼ 0 expressed in Eq. (8). After successful solutions are obtained, the
ði ¼ 1; 2; . . . N; j ¼ 1; 2; . . . MÞ (15) final values of load-carrying capacity and friction for each h/Rq and
g are obtained in the present study by averaging the solutions from
where Ai,j–Fi,j are coefficients whose expressions are omitted here 20 different, but statistically identical rough surfaces, since
for simplicity. computation results in the present study show that when the
Fig. 8. Comparisons of the friction coefficients between transverse or longitudinal surfaces with and without dimples at various sliding speed when roughness deviation
Rq = 0.0289 mm.
2870 F. Meng et al. / Applied Surface Science 256 (2010) 2863–2875
number of the surfaces goes beyond 15, its effect on the resulting study), if the contact between asperities occurs. The lubricant
solution is negligible. viscosity m is 0.035 N s/m2. The composite roughness standard
deviation Rq is taken to be 0.289 mm, the same as that measured in
4. Numerical results and discussion the above experiment.
The friction coefficients against the strip sliding speed and 4.1. Friction coefficient of dimpled surfaces
applied load are computed in this section at various ratio of film
thickness to roughness, h/Rq. As input parameters, the geometry Fig. 7 shows that comparisons of the friction coefficients (f),
sizes of the dimples, strip and cylinder used in the above defined in Eq. (14), against the relative sliding speed between
experiment are employed again. In the simulation, the composite dimpled and non-dimpled surfaces, where the H = h/Rq value
Poisson ratio of the strip and cylinder is 0.3, their composite varies. As shown from Fig. 7, in the case of g = 1/100 shown in the
Young’s modulus is 200 GPa, and usual idea elastic–plastic model left column of Fig. 7 or g = 100 case shown in the right column of
is employed and yield stress is 0.2 GPa. Meanwhile, the randomly Fig. 7, reduced friction coefficient for the dimpled surfaces against
generated surfaces with Peklenik number g = 1/100 and 100, the sliding speed can be both observed in the range of the smaller
separately comparable approximately to transverse and longitu- h/Rq, compared with results from non-dimpled surfaces. This trend
dinal surfaces, are superimposed to the bottom surfaces of the is similar to the variation of the friction coefficient with various
cylinder for the computation. Moreover, the strip is assumed to be strip sliding speed in the above experiment, as shown in Fig. 4,
smooth. The number of the gridpoints in the solution domain is the where h/Rq value is small due to the fact that very thin lubricant
same as that used in generating rough surfaces, i.e., 512 512. In film appears between the cylinder bottom and strip, since they are
solving Eq. (5), the pressure around the bottom end of the cylinder tightly pressed together by the normal force applied on the top end
is set as zero, and Reynolds boundary conditions are employed; the of the cylinder. It should be pointed out that only the variation
h/Rq value is taken to be a very small value (0.5 nm in the present trend of the friction coefficient between Fig. 7 and Fig. 4 is
Fig. 9. Comparisons of the friction coefficients between transverse or longitudinal surfaces with and without dimples at various load (N) when roughness deviation
Rq = 0.289 mm.
F. Meng et al. / Applied Surface Science 256 (2010) 2863–2875 2871
compared, instead of their absolute amplitude. This is due to always hold for the varied h/Rq value. For example, in Fig. 7(e) with
mainly the fact that the exact match of the conditions in the h/Rq = 1.1, the friction coefficient for the dimpled surface decreases
simulation and in the experiment is very difficult. In the case of only over the range of U > 0.6 m/min, while it will increases when
friction coefficient against the sliding speed, its variation trend of U 0.6 m/min. In Fig. 7(g) with h/Rq = 1.3, increased friction
the dimpled surface is similar to that of the non-dimpled surface, coefficient for the dimpled surfaces occurs over the whole range of
as shown in Figs. 4 and 7. Since the dimples can form oil pockets on the sliding speed. In Fig. 7(i) with h/Rq = 1.7, reduced friction
the surfaces, as stated previously, and the number of contacting coefficient for the dimpled surfaces can be again found over the
asperities will decreases, and thus a reduction in the friction whole range of the sliding speed. But the friction coefficient in
coefficient for the dimpled surface, compared with results from Fig. 7(i) is stable with sliding speed. This results from insignificant
non-dimpled surface, is understandable. But for friction measure- variation of hydrodynamic effect of oil film with sliding speed in
ment under the larger h/Rq condition, the associated experiment this case. The unusual phenomena of dimple effect in Fig. 7(e) and
has not been done due to the fact that it is difficult to keep thick (g) can also be found in literature [3] where dimple effects on the
lubricant oil distributing between the cylinder bottom and strip journal bearing friction was studied.
because of the serious flank leakage of lubricant, when the strip Further, the cause giving rise to the unusual dimple effect on
slides. Therefore, the friction computation for the larger h/Rq value friction coefficient in Fig. 7(e) and (g) is explored. Fig. 8 shows
is conducted only in the simulation form. In the case of g = 100 in comparisons of the friction coefficient between dimple and non-
the right column of Fig. 7, the friction coefficient for the dimpled dimpled transverse or longitudinal surfaces against the sliding
transverse surface still decreases with varied h/Rq value, compared speed at various h/Rq, where the composite roughness standard
with results from non-dimpled transverse surface. The phenomena deviation Rq is taken to be one of tenth of roughness standard
can also be found in a friction coefficient experiment from deviation measured in the above experiment, i.e., Rq = 0.0289 mm.
literature [16]. But the conclusion for g = 1/100 surface does not This small Rq value means that surfaces with the same Rq value are
Fig. 10. Comparisons of the friction coefficients between transverse or longitudinal surfaces with and without dimples at various load (N) when roughness deviation
Rq = 0.0289 mm.
2872 F. Meng et al. / Applied Surface Science 256 (2010) 2863–2875
smoother so that their roughness influence on friction is negligibly computation shows that reduced friction coefficient of dimpled
small. From Fig. 8(a)–(f), reduced friction coefficients can be surfaces can also occur. Comparisons of Fig. 7 with Fig. 8 show that
observed whether for g = 1/100, or g = 100 surfaces at the same h/ when Rq value decreases, the friction coefficient also decreases
Rq values, which are similar to variations of friction coefficients in under the condition of the same g and the same h/Rq values. This is
the experiment shown in Fig. 4. For other h/Rq values, the due to the fact that the lubricant flow carried by the valley of
Fig. 11. Comparisons of the friction forces between transverse or longitudinal surfaces with and without dimples at various sliding speed (m/min) when roughness deviation
Rq = 0.289 mm.
F. Meng et al. / Applied Surface Science 256 (2010) 2863–2875 2873
asperities sharply decreases with reduced Rq value, and hydrody- combined influence of the relative sliding speed between mating
namic effect of the lubricant always is dominant, thereby reduced surfaces, and the ratio of film thickness to roughness, roughness
friction for dimpled surfaces occurring. standard deviation, and surface orientation. In addition, for the
According to the above analyses, a conclusion can be drawn that small ratio of film thickness to roughness, or small roughness
the dimple effect on the friction coefficient is subjected to the standard deviation, the dimple can reduce friction coefficient of
Fig. 12. Comparisons of the friction forces between transverse or longitudinal surfaces with and without dimples at various sliding speed (m/min) when roughness deviation
Rq = 0.0289 mm.
2874 F. Meng et al. / Applied Surface Science 256 (2010) 2863–2875
Fig. 13. The number of contacting points for transverse or longitudinal surfaces with and without dimples (a) for surface with g = 1/100; (b) for surface with g = 100.
mating surfaces. Therefore, the proper choice of these variables is similar in trend to those of non-dimpled surfaces, the reduced
important in changing friction coefficient with dimple designs. friction force for dimpled surface can be found, compared with
Besides the comparison of the friction coefficient between those of the non-dimpled surface. Since the dimples serve as oil
dimpled and non-dimpled surfaces, the comparison is also pockets on the surfaces, a reduce in the friction force is
conducted in this section in the term of applied load, since a understandable. When the load becomes larger, the positive effect
given load is often applied on one of parallel surfaces. Figs. 9 and 10 of oil pocket on the dimples becomes stronger, so the more
separately present variations of the friction coefficient f for significantly reduced friction force occurs for dimpled surfaces.
dimpled and non-dimpled surfaces at various sliding speeds Meanwhile, the amplitude of decreased friction force at the same
against the applied load. The composite roughness standard sliding speed varies with h/Rq values. In addition, the sensitivity of
deviation Rq equals 0.289 mm in Fig. 9 while it is 0.0289 mm in reduced friction force to the g, can be observed in Fig. 11. This is
Fig. 10. The left columns of the two figures separately depict the mainly due to the difference in the surface characteristics.
friction coefficients at g = 1/100, while the friction coefficients at Fig. 12 further gives variations of the friction force at different
g = 100 are presented in the right columns of the two figures. It can sliding speeds when the h/Rq varies from 0.3 up to 1.3, where the
be found from Figs. 9 and 10 that the friction coefficient of dimpled composite roughness standard deviation Rq of 0.0289 mm is used
surfaces against the applied load is similar in trend to that of non- for minimizing influence of roughness in analyzing dimple
dimpled surfaces when the same surface orientation and sliding influence on friction. As shown in Fig. 12, the friction force of
speed are used. dimpled surface against the sliding speed is similar in trend to that
For the case of g = 100, the friction coefficient for dimpled of non-dimpled surfaces. However, reduced friction forces can be
surfaces decreases, compared with that of non-dimpled surfaces, found by comparison of friction force between the dimpled and
over the whole range of load. This is similar to variations of the non-dimpled surfaces. The variation trend of friction force in
measured friction coefficient due to dimple effect in Fig. 5, where Fig. 12 with Rq = 0.0289 mm is similar to those with Rq = 0.289 mm
the cylinder bottom surface is approximately longitudinal in Fig. 11.
orientation, although the variation trend of friction coefficient According to the decrease in friction force shown in Figs. 11 and
with the applied load in Fig. 9 (Fig. 10) seems different from that in 12, and the variation of friction coefficient shown in Figs. 9 and 10,
Fig. 5 because of the difficult exact match of the conditions in the a conclusion can be drawn that dimple effect on friction force is
simulation and in the experiment. different a little from their effect on friction coefficient.
For the g = 1/100 case, the reduced friction coefficients for It is worthy noting that for lower sliding speeds, the dimple
dimpled surfaces, similar to that in Fig. 5, occur only at the larger effect on the friction forces are weak, as shown in Figs. 11 and 12.
load level. But little change in the friction coefficient between This is because the lower sliding speed supplies the weak shearing
dimpled and non-dimpled surfaces with g = 1/100 appears on the effect of lubricant, thereby the friction force low. For the larger
smaller load level. sliding speed, the stronger shearing effect of lubricant results in the
The trend difference of friction coefficient between the g = 1/ larger friction force. Besides this, a decrease of friction force with
100 and g = 100 cases on the smaller load level may be due to the increasing h/Rq value can be observed in Figs. 11 and 12. This is
fact that the lubricant flow is smoother on the longitudinal surface because increasing h/Rq value results in decreasing number of the
than on the transverse surface, thereby the friction coefficient contacting point between mating surfaces, as shown in Fig. 13,
decreasing for the dimpled longitudinal surfaces. The above thereby the friction force decreasing.
phenomena of friction coefficient implies that for lowering friction From Figs. 11 and 12, it can also be found that the friction force
coefficient, dimpled surfaces with longitudinal orientation seem is different for g = 1/100 and g = 100 cases under the same other
better than dimpled transverse surfaces. conditions, which shows the variation trend of the friction force is
sensitive to the surface orientation. It should be pointed out that
4.2. Friction force of dimpled surfaces variations of the friction force against the applied load is omitted
here, because the variations can be derived from the variations of
Since friction force plays an important role in tribological friction coefficient against load, as shown in Figs. 9 and 10.
analyses, dimple influence on friction force for parallel surfaces is
also studied in the present study. Fig. 11 gives the friction force for 5. Conclusions
dimpled and non-dimpled surfaces at various h/Rq as a function of
the sliding speed, where the Rq value equals 0.289 mm, as the same The influence of rectangle dimples with flat bottom on the
to that used in the above experiment. As shown in Fig. 11, although friction of parallel surfaces was investigated with an experiment
the friction forces of dimpled surface against the sliding speed are where dimple influence on friction coefficient was first analyzed,
F. Meng et al. / Applied Surface Science 256 (2010) 2863–2875 2875
and with numerical simulations where the influence on the friction Foundation Graduate Research Fellowship Award, US. A part of
coefficient and friction force were discussed for randomly this work is also supported by the Natural Science Foundation of PR
generated Gaussian parallel surfaces with transverse and longitu- China (No. 50975297).
dinal orientations. Meanwhile, the validities of some numerical
results were qualitatively compared with those obtained from the References
experiment. Some conclusions are drawn below:
[1] A. Kovalchenko, O. Ajayi, A. Erdemir, G. Fenske, I. Etsion, The effect of laser surface
texturing on transitions in lubrication regimes during unidirectional sliding
(1) The friction force for the parallel surfaces can decrease due to contact, Tribol. Int. 38 (2005) 219–225.
the dimple effect over the range of the larger sliding speed, [2] M. Geiger, U. Popp, U. Engel, Excimer laser micro texturing of cold forging tool
larger load or smoother surfaces. For the smaller sliding speed surfaces-Influence on tool life, Cirp Ann. Manufacturing Technol. 51 (2002) 231–
234.
or the smaller load, the effect of the dimples on the friction [3] Xiaobin Lu, M.M. Khonsari, An experimental investigation of dimple effect on the
force becomes weak, and even can be ignored. Stribeck curve of journal bearings, Tribol. Int. 27 (2007) 169–176.
(2) Dimples on parallel surfaces can reduce the friction coefficient [4] I. Etsion, G. Halperin, V. Brizmer, Y. Kligerman, Experimental investigation of laser
surface textured parallel thrust bearings, Tribol. Lett. 17 (2004) 295–300.
for cases with the smaller ratio of film thickness to roughness
[5] G. Ryk, I. Etsion, Testing piston rings with partial laser surface texturing for
(h/Rq), small roughness or larger load. Moreover, longitudinal friction reduction, Wear 261 (2006) 792–796.
surfaces seem to behave better in lowering friction than [6] U. Pettersson, S. Jacobson, Influence of surface texture on boundary lubricated
transverse ones in the same conditions. For the case of the sliding contacts, Tribo. Int. 36 (2003) 857–864.
[7] I. Etsion, State of the art in laser surface texturing, Trans. ASME, J.Tribol. 127
larger ratio of film thickness to roughness (h/Rq), larger (2005) 248–253.
roughness or smaller load, the change in the friction coefficient [8] P. Andersson, J. Koskinen, S. Varjus, Y. Gerbig, H. Haefke, S. Georgiou, B. Zhmud, W.
for the parallel surfaces due to the dimple effect depends on the Buss, Microlubrication effect by laser textured steel surfaces, Wear 262 (2007)
369–379.
combined influence of the above factors. [9] C.P. Kumar, P.L. Menezes, S.V. Kailas, Role of surface texture on friction under
(3) According to the above analyses, the positive effect of dimples boundary lubricated conditions, Tribol. online 3 (2008) 12–18.
in lowering friction for parallel surfaces should be considered [10] N. Ren, T. Nanbu, Y. Yasuda, D. Zhu, Q.J. Wang, Micro-textures in concentrated-
conformal-contact lubrication: effect of distribution patterns, Tribol. Lett. 28
in dimple applications, but the proper choice of the sliding (2007) 275–285.
conditions of mating surfaces is important. [11] Q.J. Wang, D. Zhu, Virtual texturing: modeling the performance of lubricated
(4) Only the influence of rectangle dimples with flat bottom on the contacts of engineered surfaces, Trans. ASME, J. Tribol. 127 (2005) 722–728.
[12] T. Nanbu, N. Ren, Y. Yasuda, D. Zhu, Q.J. Wang, Micro-textures in concentrated-
friction of parallel surfaces was investigated in the present conformal-contact lubrication: effects of texture bottom shape and surface rela-
study, since the kind of dimples can reduce friction to greater tive motion, Tribol. Lett. 29 (2008) 241–252.
degree than those with other shapers. Therefore, the above [13] N. Patir, Effect of Surface Roughness On Partial Film Lubrication Using An Average
Flow Model Based On Numerical Simulation, Northwestern University, Evanston,
conclusions are only useful references to reasonably design or
IL, 1978.
use rectangle dimples with flat bottom. [14] Y.Z. Hu, D. Zhu, A full numerical solution to mixed lubrication in point contacts,
STLE Tribol. Trans. 122 (2000) 1–9.
[15] Y.C. Liu, J.Q. Wang, W.Z. Wang, Y.Z. Hu, D. Zhu, Effects of differential scheme and
Acknowledgment mesh density on EHL film thickness in point contacts, Trans. ASME, J. Tribol. 128
(2006) 641–653.
[16] P.L. Menezes, Kishore, S.V. Kailas, On the effect of surface texture on friction and
The authors would like to express their gratitude to Ford Motor friction and transfer layer formation-a study using Al and steel pair, Wear 265
Company, the Boeing Company, and the National Science (2008) 1655–1669.