Available online at www.academicpaper.org
Academic @ Paper
ISSN 2146-9067
International Journal of Automotive
Engineering and Technologies
Vol. 5, Issue 2, pp. 71 – 76, 2016
Original Research Article
Modeling and Predicting of Tribological Behaviour of the Automotive
Brake Pad Using Response Surface Methodology
Funda Kahraman1, Banu Sugözü2*
1 Tarsus
2 Institute
Technology Faculty, Mersin University, Mersin, Turkey
of Natural and Applied Sciences, Selcuk University, Konya, Turkey
Received 25 April 2016 Accepted 22 June 2016
Abstract
In the present study, wear and frictional behaviours of automotive brake pad sliding against GG 20 cast
iron counter face have been investigated experimentally using a pin on disc type friction tester. Wear
tests were carried out under dry conditions. A central composite design was used to describe response
and to estimate the process parameters in the model. Empirical models have been developed to predict
wear loss and friction coefficient as a function of braking pressure and sliding velocity using multiple
regression. The statistical analysis show that the braking pressure is identified as the most dominant
factor affecting wear loss, sliding velocity is identified as the most dominant factor affecting friction
coefficients.
Key Words: wear, friction coefficient, central composite design, regression model, friction
Doi: 10.18245/ijaet.87253
*Corresponding author:
Tel:
E-mail: banusugozu@gmail.com
1. Introduction
Automotive brake system plays a key role for
effective and safe brake performance related
issues such as friction force, noise, wear
resistance, and brake-induced vibration [13]. Automotive brake pads request the
friction materials with higher and stable
friction coefficient, low wear and noise, low
cost and composed of environmentally
friendly components. The brake system is
very important component for vehicles and
machinery equipment in industries [4]. The
researchers investigate wear and frictional
behaviours of brake system because of the
adverse effect observed in the performance
and life of brake components. A considerable
number of papers dealing with the wear and
frictional behaviours of automotive friction
materials have been published [5-13].
Several researchers have developed
mathematical models by using response
surface methodology, Taguchi techniques
and artificial neural networks to predict the
wear loss and the friction coefficient in terms
of various process parameters of different
materials [14-20].
Automotive friction materials have been
formulated for about 100 years [6]. In the
early 1920s, asbestos fiber was chosen as a
friction material to use in all kind vehicles.
Nowadays, non-asbestos organic formula
becomes very important to overcome the
negative effect of asbestos on human
respiratory system. The abrasives in the
brake friction materials play important roles
in determining the stopping distance, counter
disc wear and noise propensity [7]. The
selection of the abrasives used in commercial
brake friction materials depends on their
hardness, size, shape, fracture toughness,
wear resistance and aggressiveness against
the counter discs [9].
In this study, the tribological performance of
the non-asbestos organic type noncommercial brake pad sliding against GG 20
cast iron counter face has been investigated
experimentally under dry sliding conditions.
The Design of Experiments was done based
on response surface methodology (RSM).
Braking pressure and sliding velocity were
selected as factors, wear loss and friction
coefficient were selected as response
variables for the statistical analysis.
2. Experimental Procedures
The brake pad material studied in this study
was based on a typical non-asbestos organic
type and contained a binder, reinforcements,
fillers, and friction modifiers. The
composition of the investigated non-asbestos
friction material is summarized in Table 1.
Table 1. The ingredients of the non-commercial
brake pad materials (weight %)
The ingredients
%
Phenolic resin
15
Steel Fibers
10
Brass Particle
4
Graphite
10
Cu particles
15
Cashew
10
Barite
29
Al2O3
7
Brake pad specimens were produced by a
conventional procedure for a dry formulation
following dry-mixing, pre-forming, hot
pressing, post-curing, scratching, and
grinding. All constituents were weighed with
a sensitivity of 0.1 mg and mixed in a blender
for 3–4 min until a uniform dispersion was
obtained. After the mixing operation, the
mixture was charged in a mold. The mixture
was hot pressed at 150◦C and 15 MPa braking
pressure for 10 min and subsequently post
cured. The brake pad was produced
approximately 25.4 mm in diameter. A pin on
disc test equipment used in this study has
been shown in Figure 1. Tests with GG 20 as
the counter face material were carried out
under dry sliding condition at room
temperature. The test sample was mounted
on the hydraulic holder and pressed against
the flat surface of the rotating disc. Disc
samples of 227 mm in diameter and 9.75 mm
in thickness were obtained from a domestic
company in Turkey. Before performing the
tribological test, the surfaces of the test
samples and the GG 20 cast iron discs were
ground with 320-grid emery paper. The
average roughness (Ra) of gray cast iron disc
was measured as 1.40 µm. The hardness of
the disc samples was also measured as 191
72
HB using 5 mm ball and 150 kgf load. By
weighing the test samples to determine loss
of mass in braking pads, wear loss amount
was calculated for each test conditions.
a
b
avoid accumulation of errors. The
experiments were conducted based on
randomised run number.
Three replications of each factor level
combinations were conducted resulting in a
total of 39 tests and average values were
reported. To apply different abrasive
conditions during each test, on the rotating
disc surface and the sample is fixed in a
holder. The samples were loaded against the
abrasive medium. Samples were weighed by
analytical scales with 0.01 g sensitiveness.
After each test, samples were weighed again.
The wear loss was computed from the mass
loss of the sample. The friction coefficient
values were obtained from databank in wear
tester machine.
3. Results and Discussion
Figure 1. Test set up a) Schematic illustration pin
on disc wear tester, b) Wear test machine
Table 1. Experimental factors and levels for CCD
Factors/Levels
-1.41
-1
0
1
+1.41
Braking pressure (kPa)
438
500
650
800
862
Sliding velocity (m/s)
5.4
6
7.5
9
9.6
The objective of the study is to find the
influence of factors affecting the tribological
performance of brake pad-GG 20 disc
system. Braking pressure and sliding velocity
were considered as model variables and wear
loss and friction coefficient as response
variables. The Design Expert 7 software was
used for designing and analysing experiment.
Central composite design (CCD) was
adopted to obtain an empirical model of wear
loss and friction coefficient as a function of
the braking pressure (kPa) and sliding
velocity (m/s). The range of each parameter
was coded in five levels (−1.41, −1, 0, 1,
+1.41). The levels of model variables were
shown in Table 2. The plan of experiments
was made by randomising the experiments to
CCD is an experimental design in RSM for
building an empirical model for the response
variable. This design consists of a factorial
portion and axial portion and a central point
[21]. Experimental levels for process
variables were selected according to a CCD.
This design has 9 different design points for
all combinations of process variables. The
arrangement and the experimental (actual)
and predicted wear loss values of based on
the CCD rotatable design are shown in Table
3. The experimental results have been used to
build the mathematical models. Regression
analysis indicate second order model
adequately represents the wear loss and
friction coefficient. In order to predict, wear
loss and friction coefficient second order
regression equations were expressed in
Equations 1 and 2 in terms of coded factors.
y 0.094 0.067 x1 0.022 x 2 0.038x1 0.016x 2
2
2
(1)
0.30 0.037x1 0.10x2 0.026x12 0.012x22 0.012x1x2
(2)
where; y is the estimated wear loss, µ is the
friction coefficient, x1 is the coded factor that
represents the braking pressure x2 is coded
factor represents sliding velocity.
Analysis of Variance (ANOVA) tables for
the second order model for the wear loss and
friction coefficient have been given in Table
4 and Table 5, respectively. ANOVA was
used to identify the relationships between the
73
output and input parameters. It was also
employed to find significance of the factor
effects based on a significance level 5%.
When the p value is less than 0.05, the results
indicate statistical significance. The
performance criterion such as mean absolute
percentage error (MAPE) and absolute
fraction of variance (R2) were used to the
developed models. The value of Adjusted R2
between experimental results and predictive
values is obtained 97% for the wear loss. The
R2 value indicates that the wear parameters
explain 97% of variance in wear loss. This
value showed that the empirical model fits
well with experimental results. The
comparisons of experimental results with the
CCD predictions have been depicted in terms
of MAPE. It was found to be as 3% for the
wear loss. The value of Adjusted R2 between
experimental results and predictive values is
obtained 96% and MAPE was found to be as
5.3% for the friction coefficient. Results in
Table 4 suggest that the most significant
effect on the wear loss was exhibited by the
braking pressure, followed by sliding
velocity. Results in Table 5 suggest that the
most significant effect on the friction
coefficient was exhibited by the sliding
velocity, followed by the braking pressure.
As seen Tables 4 and 5, higher F value
indicates that the variation of the process
parameter makes big changes and
demonstrates greatest contribution on the
wear loss and friction coefficient.
Table 3. Actual and predicted wear loss (g) and friction coefficient values for CCD
Std
Run
13
10
5
6
12
4
1
11
7
3
2
9
8
1
2
3
4
5
6
7
8
9
10
11
12
13
Braking
pressure (kPa)
650
650
438
862
650
800
500
650
650
500
800
650
650
Sliding velocity
(m/s)
7.5
7.5
7.5
7.5
7.5
9.0
6.0
7.5
5.4
9.0
6.0
7.5
9.6
Actual wear
loss (g)
0.093
0.095
0.075
0.253
0.094
0.238
0.070
0.093
0.080
0.098
0.211
0.097
0.163
Predicted
wear loss (g)
0.094
0.094
0.076
0.260
0.094
0.240
0.060
0.094
0.097
0.100
0.190
0.094
0.160
Actual friction
coefficient
0.292
0.300
0.325
0.187
0.280
0.115
0.378
0.237
0.416
0.192
0.350
0.350
0.150
Predicted friction
coefficient
0.30
0.30
0.30
0.20
0.30
0.11
0.39
0.30
0.42
0.21
0.34
0.30
0.14
Table 4. The ANOVA table for the wear loss
Sourc
e
Model
A
B
A2
B2
SS
DF
MS
F value
p- value
0.050
0.035
3.71E-003
9.87E-003
1.87E-003
5
1
1
1
1
0.010
0.035
3.71E-003
9.87E-003
1.87E-003
80.54
285.70
29.91
79.52
15.11
<0.0001 significant
0.0001*
0.0009*
0.0001*
0.0060*
Source
Model
A
B
A2
B2
AB
Table 5. The ANOVA table for the friction coefficient
SS
DF
MS
F value p- value
0.096
5
0.019
59.42
<0.0001 significant
0.011
1
0.011
34.35
0.0006*
0.079
1
0.079
244.60
<0.0001*
4.73E-003
1
4.73E-003
14.56
0.0066*
1.05E-003
1
1.05E-003
3.25
0.1142
6.00E-003
1
6.00E-003
1.85
0.2161
4. Conclusions
From the above experiments and analysis
following conclusions are derived.
Regression analysis indicate that the
second order regression model
adequately represent wear loss and
friction coefficient in terms of
process variables.
The correlation with experimental
results and predicted values was good
within the range of their investigation
74
and the individual operating
parameters.
According to the statistical analysis,
Adjusted R2 values were obtained as
0.97 and 0.96 for the wear loss and
friction coefficient, respectively.
Results show that the predicting
values close to the actual values.
Mean absolute percentage errors for
CCD were found to be as 3% and
5.3% for wear loss and friction
coefficient, respectively.
ANOVA results show that braking
pressure is the most dominant factor
affecting wear loss, and sliding
velocity is the most dominant factor
affecting friction coefficient.
5. References
[1] S. K. Mehra: A review of experimental
study of brake pad material, International
Journal of Emerging Trends in Engineering
and Development 6, 3, pp. 252-256, 2013.
[2] D. Chan, G. W. Stachowiak: Review of
automotive brake friction materials,
Proceedings of the Institution of Mechanical
Engineers 218, pp. 953–966, 2004.
[3] W. Osterle, I. Urban: Friction layers and
friction films on PMC brake pads, Wear 257,
pp. 215–226, 2004.
[4] K. W. Liew, U. Nirmal: Frictional
performance evaluation of newly designed
brake pad materials, Materials and Design
48, pp. 25-33, 2013.
[5] S. K. Rhee, M. G. Jacko, P. H. S. Tsang:
The role of friction film in friction, wear and
noise of automotive brakes, Wear 146, pp.
89–97, 1991.
[6] Y. Liu, Z. Fan, H. Ma, Y. Tan, J. Qiao:
Application of nano powder rubber in friction
materials, Wear 261, pp. 225-229, 2006.
[7] P. J. Blau, J. C. Mclaughlin: Effects water
films and sliding speed on the frictional
behavior of truck disc brake material,
Tribology International 36, pp. 709-715,
2003.
[8] B. K. Satapathy, J. Bijwe: Wear data
analysis of friction materials to investigate
the simultaneous influence of operating
parameters and compositions, Wear 256, pp.
797–804, 2004.
[9] S. J. Kim, H. Jang: Friction and wear of
friction materials containing two different
phenolic resins reinforced with aramid pulp,
Tribology International 33, pp. 477–484,
2000.
[10] I. Mutlu, C. Oner, F. Findik: Boric acid
effect in phenolic composites on tribological
properties in brake linings, Materials and
Design 28, pp. 480–487, 2007.
[11] K. W. Hee, P. Filip: Performance of
ceramic enhanced phenolic matrix brake
lining materials for automotive brake linings,
Wear 259, pp. 1088–1096, 2005.
[12] N. S. M. El-Tayeb, K.W. Liew: On the
dry and wet sliding performance of
potentially new frictional brake pad materials
for automotive industry, Wear 266, pp. 275–
287, 2009.
[13] M. H Cho, K. H. Cho, S. J. Kim, D. H.
Kim, H. Jang: The role of transfer layers on
friction characteristics in the sliding interface
between friction materials against gray iron
brake discs, Tribology Letters 20, pp. 101–
108, 2005.
[14] A. Sagbas, F. Kahraman, U. Esme:
Modeling and predicting abrasive wear
behaviour of poly oxy methylenes using
response surface methodology and neural
networks, Metalurgija 48, 2, pp. 117-121,
2009.
[15] N. S. M. El-Tayeb, T. C. Yap, V. C.
Venkatesh, P. V. Brevern: Modeling of
cryogenic frictional behaviour of titanium
alloys using response surface methodology
approach, Materials and Design 30, pp. 40234034, 2009.
[16] S. Basavarajappa, G. Chandramohan, J.
Paulo Davim: Application of Taguchi
techniques to study dry sliding wear
behaviour of metal matrix composites,
Materials and Design 28, pp. 1393-1398,
2007.
[17] B. P. Chang , H. M. Akil, R. B. Nasir, A.
Khan: Optimization on wear performance of
UHMWPE composites using response
surface
methodology,
Tribology
International 88, pp. 252–262, 2015.
[18] F. Kahraman, K. B. Sugözü: Predicting
and analyzing abrasive wear loss of poly oxy
75
methylenes
using
response
surface
methodolgy, International Participation III.
Ege Composite Materials Symposium,
Aydın, Turkey, 2015.
[19] H. Zhang, Z. Li, Z. Chen: Application of
grey modeling method to fitting and
forecasting wear trend of marine diesel
engines, Tribology International 36, 10, pp.
753–756, 2003.
[20] S.C. Vettivel, N. Selvakumar, N. Leema:
Experimental and prediction of sintered Cu–
W composite by using artificial neural
networks, Materials and Design 45, pp. 323–
335, 2013.
[21] R. H. Myers, D. C. Montgomery, C. M.
Anderson-Cook.
Response
Surface
Methodology:
Process
and
Product
Optimization Using Designed Experiments,
ISBN 1118210476, 9781118210475, John
Wiley & Sons Inc., New York, 2002.
76