Sie sind auf Seite 1von 4

CIRP Annals - Manufacturing Technology 58 (2009) 291294

Contents lists available at ScienceDirect

CIRP Annals - Manufacturing Technology


jou rnal homep age : ht t p: // ees .e lse vi er. com/ci rp/ def a ult . asp

Characterization and modelling of the grinding process of metal matrix


composites
A. Di Ilio a,*, A. Paoletti a, D. DAddona (2)b
a
b

Department DIMEG, University of LAquila, Monteluco di Roio, LAquila, Italy


Department of Materials and Production Engineering, University of Naples Federico II, Naples, Italy

A R T I C L E I N F O

A B S T R A C T

Keywords:
Grinding
In-process measurement
Metal matrix composite

In this paper, some relationships for modelling force components, cutting energy and workpiece surface
roughness in grinding of metal matrix composites, are proposed. To this end, experimental data obtained
from tests carried out on a horizontal surface grinder are employed. Grinding wheels based upon alumina
abrasive were used and aluminium alloy samples reinforced with silicon carbide were ground. The
empirical models obtained could be utilised to predict how the cutting parameters affect the grinding
process and the machining quality of these non-traditional materials. The inuence of shape, orientation
and content of the reinforcement on material grindability is taken into consideration.
2009 CIRP.

1. Introduction

2. Experimental tests

The increasing use of metal matrix composites (MMCs) in


several industrial applications is due to their high specic strength
and stiffness, as well as high temperature resistance. In order to
make these applications possible, MMC components need to be
formed into the desired shapes and nished to the required
dimensions and tolerances. Metal matrix composites are given
their required shape by brazing, bonding, powder metallurgy
techniques, casting, metal spraying and by forming operations
such as bending, swaging, drawing and extrusion. Although
advances have been made in near-net-shape technology, nishing
operations are often required to obtain dimensional tolerance as
well as good surface nish. Machining of these new materials
requires tool materials of very high wear resistance because the
reinforcement is extremely abrasive [1]. Among traditional
machining processes, grinding operation is important for MMCs,
since it could be applied also in heavy-duty machining, in addition
for obtaining desired dimensional tolerances and surface quality.
Up to the present, a lot of work has been carried out to understand
the mechanisms of grinding conventional materials by analysing
the process as an interaction system between the surface of the
wheel and the workpiece [2,3]. On the contrary, there are only few
investigations on the grindability of metal matrix composites,
most of them concerning the role of reinforcement and the
inuence of grinding wheel abrasive on the process [48]. Since
grinding is a complex manufacturing process with a lot of
quantities which inuence each other, modelling can be a useful
tool to the comprehension and simulation of the process itself. In
this work some empirical relationships for predicting how the
grinding parameters affect the grinding force, the specic grinding
energy and the workpiece surface roughness are presented.

Experimental tests were carried out on a grinding machine, using


a grinding wheel based upon alumina abrasive, type 32A 46-IV. The
cutting parameters adopted were: wheel peripheral speed,
ys 2022 m=s; depth of cut, ae = 0.010.09 mm; workpiece speed,
yw 150310 mm=s. The experimental apparatus consists of a
horizontal spindle surface grinder equipped with a two components
piezoelectric dynamometer, clamped to the grinding table. A coupled
charge device (CCD)-camera system was placed on the head of the
grinding machine in order to obtain an on-line assessment of
progressive tool wear. A mechanical prolometer was employed to
measure the workpiece surface roughness off-line. The decreasing
cutting ability of the wheel during the grinding of MMCs may be
caused by the following phenomena: (1) break out and fragmentation of grains due to abrasion of reinforcement; (2) attrition wear of
the active grains; (3) clogging of the wheel caused by the adherence
of the chips. The last two forms of damage determine the formation
of wear ats on the wheel surface [5,6]. These at worn areas form
highly reecting zones on the wheel active surface, which can be
detected by means of a CCD-camera, provided with a perpendicularly
incident light beam. In fact, in this condition, the wear ats reect the
light perpendicularly, while other zones of the wheel surface produce
the scattering of light, thus reducing the light fraction which returns
to the sensor. The materials employed for tests are aluminium alloys
series 2009 and 6061 reinforced with different volume percentage of
silicon carbide in form of powders and whiskers. The code Al-xxxx/
SiC-yyR denotes an aluminium alloy series xxxx, containing yy
% volume content of reinforcement in form of Powders (R replaced
with P) or Whiskers (R replaced with W). All experimental data were
obtained adopting dry conditions and constant dressing parameters
of the grinding wheel. Consequently, the proposed models do not
take the inuence of grinding uid and dressing conditions into
consideration. For all experimental runs, almost three replicates
were carried out and the mean value of each set of data was
employed in order to process empirical relationships.

* Corresponding author. Tel.: +39 0862 434318; fax: +39 0862 434303.
E-mail address: antoniomaria.diilio@univaq.it (A. Di Ilio).
0007-8506/$ see front matter 2009 CIRP.
doi:10.1016/j.cirp.2009.03.023

A. Di Ilio et al. / CIRP Annals - Manufacturing Technology 58 (2009) 291294

292

3. Modelling

Table 1
Coefcients of Eq. (1).

Empirical models are used in all elds of machining technology


because their development requires less effort than physical
models. However, every empirical model can be employed for the
description of only one machining application. The coefcients of
the empirical models have been determined with the aid of
regression analysis methods on the basis of a number of
experimental results. Regression analysis is the generic term for
any mathematical statistical method that aims to nd a functional
interrelation between dependent random variables (measured
data points) and one or more independent random variables. The
independent variables are represented by the input parameters of
the grinding process that are the setting parameters. The output
parameters and the machining results respectively, are the
dependent variables [3]. In this work empirical equations are
determined by carrying out regression analysis of experimental
results by using the least square method.

Material

f1

f2

f3

f4

Al-2009/SiC-20P
Al-2009/SiC-15W

43
50

94
114

67
85

4
7

3.1. Grinding force models


The relevant literature offers many approaches to this eld of
modelling. Generally, since close relationships between force
components exist, it is possible to make no distinction between
the models for describing normal forces and those describing
tangential components, as these forces can be brought into relation
to each other through the grinding force ratio. The relationship
between the normal, F 0n , and tangential, F 0t , components of the
grinding force per unit width of grinding wheel for the MMCs under
test is shown in Fig. 1. The values refer to a wheel perfectly sharp, i.e.
they have been measured in the rst pass after the wheel dressing
operation; therefore they do not take into account of the changes in
grinding wheel topography due to the progressive degradation
which occurs as the number of passes increases.
As can be seen, the relationship between the force components is
linear and the slopes of the regression equations are 1.55 for the
composite Al-2009/SiC-20P (regression coefcient, Rsq = 0.99) and
1.65 for the composite Al-2009/SiC-15W (Rsq = 0.98). The ratio of the
specic normal force to the specic tangential one keeps constant
regardless of changes of equivalent chip thickness (heq ae  yw =ys ).
This fact means that, for the composite materials under test, the
ductile behaviour of the metallic matrix is predominant on the
brittle one of the ceramic reinforcement. In fact in grinding of
ceramics, the chip formation is characterised by cracking and
propagation of cracks, resulting that the ratio of normal to tangential
force presents quite strong non-linearity. For ductile materials, the
chip formation is due to plastic deformation and it could be
considered as a stable process. The analysis of the experimental
results obtained in this work has shown that the normal and
tangential components of the specic force may be fairly well
approximated by a polynomial relationship of the equivalent chip
thickness, such as Eq. (1) relating to the normal component:
F 0n f 1  h3eq f 2  h2eq f 3  heq f 4

(1)

Fig. 2. Specic normal and tangential forces as a function of the equivalent chip
thickness.

Fig. 3. Specic normal force as a function of the specic material removal.

The values of the coefcients, determined by applying the least


square method, are reported in Table 1, while Fig. 2 shows the
resulting trends of the grinding forces as a function of the
equivalent chip thickness.
The processing of the experimental results has allowed to
hypothesize the following model which takes into account the
inuence of the cutting parameters:
n
F 0n C 1  am
e q

(2)

where q yw =ys , is the speed ratio and the values of the


coefcients C1, m and n, obtained by the regression analysis, are
shown in Table 2.
The above-mentioned force models are based on the assumption of a stationary grinding process. The inuence of the grinding
time and the resulting changes in grinding wheel topography can
be taken into account considering the specic material removal
(V 0w ). Fig. 3 shows the inuence of the specic material removal on
the trend of the specic normal force.
The non-stationary behaviour of the grinding force can be
ascribed to the increasing degradation of the wheel as the number of
passes increases. In the case of MMCs under test it has been found
that the main responsible of wheel degradation is the clogging
produced by the tough matrix [6]. The following relationship,
Table 2
Coefcients of the forces models.

Fig. 1. Specic normal force as a function of the specic tangential force.

Material

C1

Al-2009/SiC-20P
Al-2009/SiC-15W

265
593

0.79
0.86

0.19
0.28

A. Di Ilio et al. / CIRP Annals - Manufacturing Technology 58 (2009) 291294

293

determined by a regression analysis of the present experimental


results, shows a force model which takes the grinding time into
consideration during machining of Al-2009/SiC-20P composite:
F 0n 535  a0:72  q0:36  V 0w 0:12

(3)

It has been found in the investigation, that the grinding force reduces
with increasing the material hardness, which in turn depends on the
volume content of reinforcement. However, since the hardness also
depends on dimensions and distribution of the particles inside the
matrix, it is difcult to establish a clear relationship between the
grinding forces and the volume content of reinforcement.
3.2. Grinding energy models
The energy spent during grinding causes the heating of the
ground surface, which leads to the risk of material damage. The
specic grinding energy, U, dened as F 0t  ys =ae  yw , can be
considered to consist of cutting, which includes chip formation and
plowing contributions, and sliding components, according to the
following formula [9]:
U U ch U plowing U sliding

(4)

In Fig. 4, the trends of specic grinding energy as a function of


removal rate per unit width are reported.
Experimental results may be interpolated by curves exhibiting
the following structure:
U C 2 ae  yw p

(5)

The coefcients C2 and p assume the values 60 and 0.61,


respectively, for the composite Al-2009/SiC-20P, and 57 and 0.49,
respectively, for the composite Al-2009/SiC-15W. At low removal
rates, the specic grinding energy is quite large, while it decreases
at higher removal rates tending towards a limit value which
depends on the workpiece material. As it is known, the specic
tangential cutting force F 0t;c is equivalent to that of a perfectly sharp
wheel, where the at area fraction is equal to zero. The evaluation
of F 0t;c requires the development of a force trend as a function of
wear at area percentage, A, as shown in Fig. 5 for the composite
Al-2009/SiC-15W.
In this diagram, the intercepts of the ordinate axis allow to
evaluate the values of the specic tangential cutting force, in
accordance with the method proposed by Malkin [9]. As can be
seen from Fig. 5, for the MMCs under test, the tangential force is
almost unaffected by the wheel degradation, which is mainly
caused by clogging of adhered chips. On the contrary, the normal
force is markedly affected by the wheel degradation, as shown in
Fig. 3, therefore the proportionality between F 0n and F 0t shown in
Fig. 1 is lost as the grinding wheel becomes dull. The trends
reported in Fig. 5 evidence that the contribution of sliding to the
tangential force, and hence to the energy expended for grinding, is
minimum. However, in order to estimate more accurately the
fraction of tangential force due to chip formation, F 0t;c , the
coefcients of the linear regression equations for the trends of

Fig. 4. Specic grinding energy as a function of the removal rate per unit width.

Fig. 5. Specic tangential force as a function of the at area percentage.


Table 3
Coefcients of the regression lines of the tangential force as a function of the at
area percentage.
Q 0w (mm3/mm s)

Slope

Intercept, F 0t;c

7.5
10.5
15.0
27.0

0.07
0.09
0.19
0.17

8.17
9.31
9.75
15.6

tangential force as a function of at area percentage have been


calculated and reported in Table 3.
The values of F 0t;c reported in Table 3 have been used to calculate
the specic grinding energy for cutting, Uc, dened as
Uc

F 0t;c  ys
;
ae  yw

which is reported in Fig. 6 as a function of the material removal rate.


From the results shown in Fig. 6, the specic chip formation
energy, Uch, which is actually expended by material removal, can
be approximately estimated to be equal to 10 J/mm3.
3.3. Surface roughness models
In MMCs grinding, the morphology of the ground surfaces is
characterised by the presence of side ow ploughing marks and
scratches and by areas which evidence high plastic deformation
and lack of ridges. It has been noted that the zones with high plastic
deformation decrease for materials which exhibit higher hardness
values. Similarly to conventional materials grinding, one starting
point for empirically relating surface roughness to the grinding
parameters is to assume a direct correlation with the equivalent
chip thickness, which, as suggested in ref. [9], assumes the form:
Ra C 3  hxeq

(6)

where Ra is the average roughness, and C3 and x are coefcients, the


values of which were determined by applying a regression analysis
of the experimental data and are shown in Table 4. Fig. 7 reports in

Fig. 6. Specic grinding energy for cutting as a function of the removal rate per unit
width.

A. Di Ilio et al. / CIRP Annals - Manufacturing Technology 58 (2009) 291294

294

Table 4
Values of the constants for surface roughness models.
Material

C3

C4

Al-2009/SiC-20P
Al-2009/SiC-15W

0.69
1.36

0.16
0.27

2.63
15.2

0.15
0.26

0.22
0.40

Fig. 9. Surface roughness as a function of the specic tangential force.

The regression equation which links the average roughness to


the Rockwell B hardness is given by:
Ra 0:007  HRB 1:16
Fig. 7. Surface roughness as a function of the equivalent chip thickness.

a semi-logarithmic plot the surface roughness trend as a function


of the equivalent chip thickness for the two composites under test.
As can be noted, the composite reinforced with powder exhibits
a better surface nish than that reinforced with whiskers,
nevertheless both composites have almost comparable hardness
values. In fact, the Rockwell Hardness B (HRB) is equal to 67.6  1.5
for the composite Al-2009/SiC-20P and to 69.5  1.5 for the
composite Al-2009/SiC-15W.
The differences in roughness values can be ascribed to a lower
notch effect associated with the roundness of the powder. Eq. (6)
assumes that the inuence of depth of cut and workpiece speed
have the same effect on roughness. Indeed, it has been found that
the roughness depends in a different manner on these machining
parameters, therefore the regression using the following equation
was performed:
Ra C 4  aye  qz

(7)

where C4, y and z are coefcients, the values of which are reported
in Table 4.
Also for roughness, the inuence of grinding time can be taken
into account by considering the specic material removal, through
the following empirical equation:
Ra 7:66  a0:13
 q0:36  V 0w 0:09
e

(8)

where the coefcients have been determined for Al-2009/SiC-15W


composite. As can be seen, the inuence of the grinding time is
almost negligible with respect to that of the cutting parameters.
For the composite reinforced with powder, the surface roughness
has shown a decreasing linear trend as a function of the hardness of
the workpiece material, as evidenced in Fig. 8.

Fig. 8. Surface roughness as a function of the workpiece hardness.

(9)

As both grinding force components and surface roughness are


inversely related to the material hardness, it is possible to establish
a correlation between surface roughness and specic tangential
force, as reported in Fig. 9.
The resulting regression equation is:
Ra C 5  F 0t C 6

(10)

where C5 = 0.28 and C6 = 0.43 for the composite Al-2009/SiC-15W


and C5 = 0.41 and C6 = 0.25 for the composite Al-2009/SiC-20P. The
above data have been obtained for different values of the specic
material removal, while constant grinding parameters have been
adopted. Consequently, the diagram also represents the trend of the
surface roughness as a function of the specic grinding energy.
4. Conclusions
An investigation on the modelling of the grinding process of
MMCs, based on empirical relationships, was proposed. Empirical
models can help to understand the process and grinding xture
design. The conclusions obtained can be summarised as follows:
(1) the relationship between the normal and the tangential
components of grinding force is linear; (2) the sliding component
of the specic grinding energy is almost negligible with respect to
the cutting component, which shows a decreasing exponential
trend as the removal rate increases; (3) the workpiece surface
roughness can be related with the equivalent chip thickness
through a power relationship; it shows a decreasing linear trend as
the hardness of workpiece material increases.

References
[1] Cronjager L, Meister D (1992) Machining of Fibre and Particle-reinforced
Aluminium. Annals of the CIRP 41(1):6366.
[2] Tonshoff HK, Peters J, Inasaki I, Paul T (1992) Modelling and Simulation of
Grinding Processes. Annals of the CIRP 41(2):677688.
[3] Brinksmeier E, Aurich JC, Govekar E, Heinzel C, Hoffmeister HW, Klocke F, Peters J,
Rentsch R, Stephenson DJ, Uhlmann E, Weinert K, Wittmann M (2006) Advances in
Modeling and Simulation of Grinding Processes. Annals of the CIRP 55:667696.
[4] Zhong Z, Hung NP (2002) Grinding of Alumina/Aluminium Composites. Journal
of Materials Processing Technology 123:1319.
[5] Zhu Y, Kishawy HA (2004) Inuence of Alumina Particles on the Mechanics of
Machining Metal Matrix Composites. International Journal of Machine Tools &
Manufacture 45:389398.
[6] Di Ilio A, Paoletti A, Tagliaferri V, Veniali F (1996) An Experimental Study on
Grinding of Silicon Carbide Reinforced Aluminium Alloys. International Journal
of Machine Tools & Manufacture 36/6:673685.
[7] Di Ilio A, Paoletti A (2000) A Comparison Between Conventional Abrasives and
Superabrasives in Grinding of SiC-Aluminium Composites. International Journal
of Machine Tools & Manufacture 40(2):173184.
[8] Kwak JS, Kim YS (2008) Mechanical Properties and Grinding Performance on
Aluminium-Based Metal Matrix Composites. Journal of Materials Processing
Technology 201:596600.
[9] Malkin S (1989) Grinding TechnologyTheory Applications of Machining with
Abrasives. Ellis Horwood Limited, West Sussex.