Int. J. Mechatronics and Manufacturing Systems, Vol. 4, Nos.

3/4, 2011 337

Design of an active control of vibration in a

centreless grinding machine: theoretical study and
experimental implementation

Iker Garitaonandia
Department of Mechanical Engineering,
Faculty of Mining and Civil Engineering,
University of the Basque Country, UPV-EHU,
Colina de Beurko s/n, 48902, Barakaldo, Bizkaia, Spain
E-mail: iker.garitaonandia@ehu.es

Joseba Albizuri
Department of Mechanical Engineering,
Faculty of Engineering,
University of the Basque Country, UPV-EHU,
Alameda de Urquijo s/n 48013, Bilbao, Bizkaia, Spain
E-mail: joseba.albizuri@ehu.es

María Helena Fernandes*,

Jesús M. Hernández and Itxaso Olabarrieta
Department of Mechanical Engineering,
Faculty of Mining and Civil Engineering,
University of the Basque Country, UPV-EHU,
Colina de Beurko s/n, 48902, Barakaldo, Bizkaia, Spain
E-mail: mariahelena.fernandes@ehu.es
E-mail: jesusmaria.hernandez@ehu.es
E-mail: itxaso.olabarrieta@ehu.es
*Corresponding author

David Barrenetxea
Department of Manufacturing Processes,
IDEKO IK4, Pol. Industrial de Arriaga,
2. 20870, Elgoibar, Spain
E-mail: dbarrenetxea@ideko.es

Abstract: The appearance of vibrations during centreless grinding process is

one of the main factors limiting the current requirements of precision and
productivity. This paper presents a novel procedure for the design of an active
vibration reduction system in a centreless grinding machine using piezoelectric
actuators. The different phases involved in the modelling procedure of the
machine tool are explained, together with the followed approach to locate the
actuators. Taking as reference the finite element model of the machine, the

Copyright © 2011 Inderscience Enterprises Ltd.

338 I. Garitaonandia et al.

most important mode shapes affecting the chatter behaviour have been
considered to obtain a low order state space model. This reduced model has
been used to simulate the effectiveness of the control strategy and the results
have been compared with those obtained experimentally after the system
implementation. A good agreement has been obtained demonstrating that the
modelling procedure followed is useful to predict the effectiveness of control
systems.

Keywords: centreless grinding; active control; vibration reduction.

Reference to this paper should be made as follows: Garitaonandia, I.,

Albizuri, J., Fernandes, M.H., Hernández, J.M., Olabarrieta, I. and
Barrenetxea, D. (2011) ‘Design of an active control of vibration in a centreless
grinding machine: theoretical study and experimental implementation’,
Int. J. Mechatronics and Manufacturing Systems, Vol. 4, Nos. 3/4, pp.337–355.

Biographical notes: Iker Garitaonandia is an Assistant Professor in the

Department of Mechanical Engineering of the University of the Basque
Country (UPVEHU). He obtained his PhD in Industrial Engineering from the
University of the Basque Country in 2009. His main fields of interest are:
mechanics of materials, continuum mechanics, machine vibration control, and
active materials.

Joseba Albizuri is a Professor in the Department of Mechanical Engineering of

the University of the Basque Country (UPV-EHU). He obtained his PhD in
Industrial Engineering from the University of the Basque Country in 1994. His
main fields of interest are: modelling of machines and structures by FEM,
machine dynamics, experimental modal analysis, and technology of materials.

M. Helena Fernandes is a Professor in the Department of Mechanical

Engineering of the University of the Basque Country (UPV-EHU). She
obtained her PhD in Industrial Engineering from the University of the Basque
Country in 1994. Her main fields of interest are: modelling of machines and
structures, experimental modal analysis, and experimental techniques.

Jesús M. Hernández is a Professor in the Department of Mechanical

Engineering of the University of the Basque Country (UPV-EHU). He obtained
his PhD in Industrial Engineering from the University of the Basque Country in
1994. His main fields of interest are: modelling of machines and structures,
experimental modal analysis, and updating of finite element models.
Itxaso Olabarrieta is an Assistant Professor in the Department of Mechanical
Engineering of the University of the Basque Country (UPV-EHU). She is in her
second year of her MS in Mechanical Design at the University of the Basque
Country. Her main fields of interest are: modelling of machines tools, model
and updating techniques.
David Barrenetxea is a PhD candidate in Mechanical Engineering at University
of Mondragon (MU). He is a Research Engineer working in the fields of
analysis of machine tool behaviour based on the finite element method and
theoretical and experimental dynamics. His main fields of interest are: analysis
and optimisation of grinding process behaviour by means of modelling and
simulation techniques, and machine tool design.
 Design of an active control of vibration in a centreless grinding machine 339

1 Introduction

Among the grinding processes, centreless grinding is especially important in many

industrial sectors that require mass-production, because it allows to obtain high removal
capacities with an excellent precision and surface finish. One of its main advantages is
that it allows loading and unloading of the pieces in an extremely easy and quick way.
Usually, this process is used to machine cylindrical pieces with high surface finish
requirement.
During the centreless grinding process, the appearance of vibrations, especially the
regenerative chatter, limits to reach the demands of precision and productivity. For this
reason, different researchers have dedicated efforts to find solutions to this problem.
Basically, the procedures used have been: the adequate selection of setup conditions, the
structural modification of the machine and the use of active vibration control strategies.
Several authors, after developing stability diagrams, proposed different combinations
of parameters to obtain configurations free of chatter (Miyashita, 1972; Miyashita et al.,
1982; Hashimoto et al., 2000; Hashimoto and Lahoti, 2004). These diagrams show two
stable areas of operation, one at low angular velocities of the workpiece and another one
at high angular velocities.
Zhou et al. (1996) and Zhou (1994) developed stability diagrams that enabled to
deduce relationships between grinding geometrical parameters and the origin of the lobes
in the workpiece surface.
Furukawa et al. (1973) presented a procedure to modify the angular velocity of the
wheels when the chatter vibrations appear. Taking account of the same idea, Zhou (1994)
intended to regulate adequately the angular velocity of the regulating wheel to maximise
the interference generated among successive waves of the piece, so that the produced
roundness error was minimised.
Static and dynamic stiffness have an important influence on the precision of the
machine tools because an increment of stiffness restricts the structural deformations
produced by the cutting forces. With the idea of acting on the stiffness of the machine,
Rowe et al. (1987) designed and developed a centreless grinding machine prototype
where the spindles were in-line with the force transmission axis. In this way, high
bending strains of the wheel heads, which produce chatter vibration, were avoided. By
means of an appropriate selection of the geometric configuration of the machine, an
appreciable reduction of roundness errors of the machining pieces was reached.
Also, Hashimoto et al. (2000) developed a modified centreless grinding machine,
whose first vibration frequency was increased to 385 Hz. They were able to increase the
static stiffness of the machine and improve the stability of the process for low angular
velocities of the workpieces.
With the idea of improving dynamic stiffness in the centreless grinding process,
Bennett et al. (1994) carried out a comparison between three types of slide-way designs
between upper slide and lower slide, and between lower slide and swivel plate. The first
system corresponded to the dovetail system, traditionally employed in the machine tool
design. The second design corresponded to a system based on the recirculation of rolling
elements together with dampers. In the third, they used a cover of polymer material on
the contact surfaces of the dovetails between the mentioned components. From the
slide-way system models, the numerical frequency response characteristics were
obtained. It was concluded that the dynamic behaviour of the conventional system could
340 I. Garitaonandia et al.

be improved if an adequate stiffness in the second design was used and if damping
mechanisms were introduced into the third one.
In 1985, in order to combine high dynamics and positioning accuracy, Yoshioka et al.
(1985) designed a composite hydrodynamic/hydrostatic bearing slide-way.
As for the use of control techniques, the first active control system of vibrations in a
machine tool was proposed by Comstock et al. (1969) for the turning process.
Afterwards, Nachtigal and Cook (1970) designed a control system for turning based on a
feedforward loop, where the cutting forces during machining were measured and used to
predict the displacement of the cutting tool.
Mitchell and Harrison (1977) were the first authors using the observer theory to
design an active controller in machine tools. They implemented a predictive controller in
a turning machine based on closed-loop feedforward strategy.
As for active vibration control applications in grinding, Gosebruch (1987) presented
an active rigidisation methodology for the clamp elements of pieces in external
cylindrical grinding. Gao et al. (2001) presented a micropositioning system using a
piezoelectric actuator. Altintas and Weck (2004) showed the application of an active
control system of vibrations in the cylindrical grinding.
In centreless grinding machines, Albizuri et al. (2007) implemented an active
feedback strategy using piezoelectric actuators.
In this paper, the design process of an active control system in order to reduce
vibrations in a centreless grinding machine is presented. This process includes machine
modelling, selection of the location of the actuators, model reduction, prediction of the
behaviour by means of reduced models, and practical implementation of the system.

2 Characteristics of the centreless grinding machine

The studies shown in this paper were carried out in a centreless grinding machine, model
ESTARTA-327 MDA. In Table 1, technical characteristics of the machine are presented.
Table 1 Technical characteristics

Value Units
Net machine mass: 14,000 kg
Grinding range ∅: 1.5 to 200 mm
Max. length of grinding in plunge: 500 mm
Grinding wheel (max.): ∅ 650 × 500 mm
Regulating wheel (max.): ∅ 355 × 500 mm
Peripheral speed grinding wheel: 35 m/s
Regulating wheel speed in operation: 10–225 min–1

In order to identify what vibration mode shapes are excited in chatter presence, several
machining tests were performed under chatter conditions.
In these tests, the structural response was measured in different locations of the
machine using a triaxial accelerometer. From the analysis of these signals, it was
concluded that the most important vibrations take place in the infeed direction of
the upper slide (Z direction in Figure 3 and Figure 4). The frequency content of the
 Design of an active control of vibration in a centreless grinding machine 341

acceleration signals of both grinding wheel head and workblade are shown in Figure 1
and Figure 2.
The wheel vibrations were dominated by a vibration mode which was excited around
55 Hz. On the other hand, the workblade vibrations were dominated by a mode around
120 Hz, which produced a local bending movement of the lower slide in the joint area
between the two elements, moving the workblade in its path. Due to the great influence
that the vibrations in the wheels have in the roundness errors of workpieces, the first
mode was considered more important than the second one.

Figure 1 Frequency content of the acceleration signal at the grinding wheel head under chatter
conditions

Figure 2 Frequency content of the acceleration signal at the workblade under chatter conditions
342 I. Garitaonandia et al.

Figure 3 Main chatter mode

Figure 4 Finite element model

Table 2 Experimental mode shapes

Mode EMA Freq. (Hz) Damp. (%)

1 33.41 5.8
2 48.43 3.53
3 58.91 3.5
4 76.97 3.9
5 90.52 2.6
6 108.44 2.6
7 122.02 1.8
8 128.48 1.6
9 129.86 1.1
10 144.02 2.0

On the other hand, an experimental modal analysis of the machine was carried out in
order to obtain the natural frequencies and mode shapes (Table 2).
 Design of an active control of vibration in a centreless grinding machine 343

The third experimental mode corresponding to the natural frequency of 58.91 Hz is

the mode identified in the machining tests that caused the most important roundness
errors in the workpieces. Due to its importance, this mode has been considered the main
mode of chatter. In Figure 3, a sequence of this mode is shown, where it is possible to
appreciate an out of phase movement between the grinding wheel head and the regulating
wheel head. Due to this movement, an important relative motion is generated between the
two wheel heads when this mode is excited, causing undulations or lobes in the
workpiece.
The ninth experimental mode corresponding to the natural frequency of 129.86 Hz is
the one that also got excited in the machining tests carried out (highest peak in Figure 2).
This mode has been considered the secondary chatter mode.

3 Machine modelling and location of the actuators

Since the centreless grinding machine is a complex mechanical system it was opted to use
the finite element method for its modelling.
In order to define the necessary characteristics of the piezoelectric actuators to be
used, the maximum amplitude vibrations were measured under chatter conditions. Tests
showed that maximum values of displacement were few microns.
The piezoelectric actuators provide a quick response for the signal voltages applied
and a great capacity of generating forces. These characteristics are indispensable to
perform an effective active control. The idea was to introduce this type of actuators inside
the passive structure in the high strain energy density (SED) zones for the main chatter
mode (Ehmann and Nordmann, 2003; Ehmann et al., 2002), since this is the zone where
they have greater control authority.
From the finite element model, the density distribution for the main chatter mode was
obtained (Figure 5). The highest concentration of modal SED was located in the element
that models the lower spindle in longitudinal direction. In consequence, this machine
zone was selected to place the actuators.

Figure 5 Distribution of the SED for the main chatter mode

max. SED

0 5 15 40 14380
1 10 20 50
344 I. Garitaonandia et al.

Two actuators P-247.30 (Physikinstrumente) were used. In Table 3, the characteristics of

these elements are shown.
Table 3 Characteristics of actuators

Characteristic Value Units

Stiffness 680 N/micron
Voltage range 0–1,000 V
Max. displacement 40 micron
Max. push force 30,000 N
Max. pull force 3,500 N
Natural frequency without load 6.5 kHz
Mass without wires 660 g

In order to introduce the behaviour of the actuators in the finite element model of the
machine it was defined the relation among the externally applied voltage (Vpzt), the axial
deformation of an actuator (ΔL) and the force exerted by the actuator (Fpzt) as:

(
Fpzt = K pzt ⋅ g a ⋅ Vpzt − ΔL ) (1)

Defining the gain of each actuator like ga = d33 ⋅ n, where d33 is the piezoelectric constant
and n is the number of piezoelectric discs.
The force Fpzt has two terms: one proportional to the applied voltage and other
proportional to the expansion. In this way, it was concluded that the modelling of these
actuators could be carried out appropriately using a stiffness Kpzt between its ends
(modelling passive behaviour) and a pair of opposite forces of value Kpzt · ga · Vpzt
applied axially in the same ends (modelling active behaviour). This equivalence is
illustrated in Figure 6.

Figure 6 Piezoelectric actuator model

This representation allows to integrate the piezoelectric actuators in the finite element
model in a simple way. The location zone was modified including actuator stiffness
(Figure 7), so the machine passive behaviour was affected. Also, it was necessary to
incorporate additional nodes at the ends of the actuators in order to represent those areas
where forces are exerted actively.
The integration of the actuators supposes a modification in the structure of the
machine that can affect its dynamic behaviour. To evaluate the model modification effect,
natural frequencies from the modified FE model were calculated in the range of
frequencies 0–160 Hz. The new frequencies and the initial ones are shown in Table 4.
 Design of an active control of vibration in a centreless grinding machine 345

The differences in natural frequencies are very small. In regard to vibration mode
shapes, for each one the modal assurance criterion (MAC) was close to 1, meaning that
the mode correlation was almost perfect. Therefore, it was concluded that the installation
of the actuators would not modify the machine passive behaviour substantially.

Figure 7 Structural modifications (see online version for colours)

Table 4 Comparison between FE frequencies before and after the structural modifications

Freq. (Hz) New freq. (Hz) Difference (%)

1 32.5 32.4 –0.31
2 33.5 32.9 –1.79
3 48.2 47.9 –0.62
4 58.9 56.2 –4.58
5 74.9 74.8 –0.13
6 86.0 85.9 –0.12
7 90.2 89.6 –0.67
8 98.3 98.5 0.20
9 104.3 103.7 –0.58
10 116.2 115.9 –0.26
11 118.3 118.0 –0.25
12 125.7 125.3 –0.32
13 139.5 139.6 0.07
14 145.6 146.5 0.62

4 The control strategy

Feedback strategies were selected in order to reduce actively the vibration levels. The
final purpose was to decrease the resonant peaks of the different vibration modes. The
general diagram for this strategy is shown in Figure 8. The signal from the control
346 I. Garitaonandia et al.

system, after going through the controller gain g, is sent to the actuators located in the
centreless grinding machine.
A second order filter (SOF) was used by means of feedback of acceleration. The
different elements that act in the control loop are represented in the modified FE model
(see Figure 9).

Figure 8 Feedback controller for active structural damping

Figure 9 SOF control (see online version for colours)

The basic idea of the SOF controller is to pass the acceleration signal through a filter of
second order, and to generate a voltage feedback proportional to the output of the filter.
Therefore, the controller shown in Figure 9, which consists on the SOF and the gain, is
described by means of the following general form:

−g ⋅ ωf2
H(s) = (2)
s 2 + 2 ⋅ ξf ⋅ ωf ⋅ s + ωf2

The controller’s poles that are defined by means of the selection of the filter properties
(natural frequency ωf, modal damping ξf), are located in the complex plane to produce an
appropriate migration of the poles of closed-loop as the gain g is increased.
In Figure 9, it is shown that the accelerometer was located close to the actuators. Due
to the physical impossibility to locate the accelerometer in the performance zone of the
actuators, the controller was not exactly collocated. But, their proximity offers guarantees
for the attainment of the unconditional stability (Preumont, 2002).
From the modified FE model, the first 15 natural frequencies were extracted
(diagonal matrix [Ωm]) while modal matrix [Φ]m was created using the modal
 Design of an active control of vibration in a centreless grinding machine 347

displacements of the first 15 vibration modes corresponding to the translational degrees

of freedoms of the model representative nodes (where application of forces or acquisition
of responses was required).
These two matrices and the modal damping matrix obtained experimentally, [ξm],
allowed to obtain a state space model of order 30 with the inputs and outputs required to
simulate the control algorithm (Garitaonandia et al., 2008).
As inputs that have influence on the machine behaviour, it should be considered,
besides the instantaneous normal grinding force, Fn(t), voltages that are supplied to the
two piezoelectric actuators coming from the output of the controller. However, the state
space model developed in this work only admits forces as inputs. Therefore, according to
the idealisation of the actuators shown in Figure 6, two forces proportional to the applied
voltages in the left and right actuators have been defined as Fleft(t) and Fright(t), according
to:

Fleft (t) = K pzt ⋅ g a ⋅ Vpzt _ left (t) (3)

Fright (t) = K pzt ⋅ g a ⋅ Vpzt _ right (t) (4)

It is interesting to highlight that according to the feedback diagram shown in Figure 9, the
controller H(s) receives an acceleration signal and provides the same voltage signal to the
two actuators. Therefore, keeping in mind that the two actuators have the same
characteristics, equations (3) and (4) show that both force Fleft(t) and Fright(t), that are
introduced in the state space model, are the same. These forces, combined with the
proportional components to the expansions of the actuators [equation (1)], provide the
total forces introduced actively to the machine.
The state space model developed allows to extract the displacements, velocities and
accelerations of the different degrees of freedom selected as outputs. In this case, besides
the variable that describes the deformation of the machine, ym(t), it was interesting to
obtain the acceleration at the degree of freedom corresponding to the accelerometer
localisation, a(t).
Therefore, the model developed has three inputs and two outputs (MIMO system).
After elaborating the state space model with the first 15 vibration modes, it is
necessary to reduce the model order considering only the most controllable and
observable modes. The procedure used to select the most important modes in the
behaviour between the different inputs and outputs was the balanced reduction
because this type of MIMO systems offers the biggest advantages (Hatch, 2001).
After transforming the state space model in the balanced representation, it was possible
to observe that the behaviour between inputs and outputs was dominated by six states of
this representation. Therefore, the reduction procedure consisted in keeping these six
states. In order to keep the system gain in steady-state, the remaining states were
considered in a residual term. After this reduction process, the state space model obtained
was:

⎧ Fn (t) ⎫
⎪ ⎪
 } = [ A r ] ⋅ {x(t)} + [ Br ] ⋅ ⎨ Fleft (t) ⎬
{x(t) (5)
⎪ ⎪
⎩Fright (t) ⎭
348 I. Garitaonandia et al.

⎧ Fn (t) ⎫
⎧ y m (t) ⎫ ⎪ ⎪
⎨ ⎬ = [ Cr ] ⋅ {x(t)} + [ D r ] ⋅ ⎨ Fleft (t) ⎬ (6)
⎩ a(t) ⎭ ⎪ ⎪
⎩Fright (t) ⎭
[Ar], [Br], [Cr] and [Dr] are the reduced matrices of the system.
Once the reduced model was obtained, the following step in the simulation process
consisted on closing the loop between the acceleration and the voltages applied to the
actuators by the control law [equation (2)]. In Figure 10, the reduced model and the
control loop are illustrated.

Figure 10 Feedback loop based on the SOF

Figure 11 FRF without control and with control

An adequate design of the feedback loop requires to adjust the frequency of the filter
poles, ωf, to the natural frequency of the mode shape of interest, using a considerable
value of the damping ratio ξf (Preumont, 2002). The frequency ωf was adjusted to
 Design of an active control of vibration in a centreless grinding machine 349

55.5 Hz, very close value to the main frequency of chatter. For the damping ratio,
ξf = 0.5 was selected.
A very important feature to consider in the simulation process is the choice of the
controller gain, as it modifies the structural dynamic behaviour, changing closed-loop
roots location. Studying the evolution of these roots by increasing values of the controller
gain, it was obtained that the root reaches a maximum damping of 22.5% for an optimum
value of the gain.
Figure 11 shows the comparison between FRF without control and the one obtained
considering the value of optimal gain.
From Figure 11, it can be concluded that the damping of the main chatter mode is
considerably increased when the controller acts. Additionally, another mode close to
33 Hz is also damped. This is a mode where all machine components move in phase in
the infeed direction in a suspension movement with respect to the supports of the bed
with the foundation. It does not substantially affect to the relative movement between the
workpiece and the wheels.

5 Time domain simulations

The simulations performed in the frequency domain have shown that the designed
controller is able to actively damp the main chatter mode, but they do not take into
account the practical limitations that can be presented when implementing the control
system. These limitations arise because the actuators do not have an infinite force
generation capacity and, in practice, the voltage level that can be supplied to them is
upper bounded (see Table 3).

Figure 12 Theoretical estimation of the vibrations in the regulating wheel head

This effect has been considered through time domain simulations. For this purpose, the
control loop shown in Figure 10 has been integrated in the chatter loop of the centreless
grinding process, following the procedure described in Fernandes et al. (2009). Some
simulations have been performed increasing gradually the controller gain of the SOF
350 I. Garitaonandia et al.

controller until the maximum allowable voltage has been estimated in the actuators. In
these conditions, the vibrations of different parts of the machine have been simulated and
compared to the vibrations estimated with no control applied. Figure 12 shows this
comparison for a point located in the regulating wheel head, where it can be seen that the
control action reduces the vibration amplitudes considerably. The study of the frequency
content of these acceleration signals has shown that the response is dominated by the
main chatter mode.

6 Experimental tests

In this section, results obtained in the experimental tests executed in the machine using
the control system based on the filter of second order are detailed. For it, the machine was
modified to incorporate the piezoelectric actuators.
To avoid tension efforts in the actuators during the machining operations, they were
preloaded. It was carried out applying a compression force to each actuator of
approximately 15,000 N (50% of compression load range).
A piezoelectric accelerometer, model Endevco 5221, was used to obtain the
acceleration signal for the control system.
The processing of the acceleration signal was implemented with a card Dspace
DS1104. The control in real-time was implemented by means of the module real-time
workshop of Matlab.
In Figure 13, the assembly used in the control implementation is presented.

Figure 13 Experimental setup

Connection panel R

A n (2 1 D
d 2
s
DS Con Filt ro Ga S DS
Filt r
D
DS
5
C

PC
Accelerometer Accelerometer
amplifier

actuators

Actuator amplifier
 Design of an active control of vibration in a centreless grinding machine 351

Initially, after amplifying the analogical accelerometer signal, this was introduced into a
PC by means of the connection panel. The signal was digitised by means of a block
retainer ADC of order 0 using a sampling time of 2 μs. This sampled input signal was
passed through a high pass filter of order 2 with a cut-off frequency of 10 Hz whose
function was to eliminate the continuous components of the input acceleration signal. The
filtered signal was used to feed the control algorithm.
The sampled signal obtained was used as input to the two DAC blocks, one for each
actuator, in order to obtain the analogical signals that feed the actuator amplifiers.
Finally, the amplifier output voltages were supplied to the piezoelectric actuators.
To evaluate the effectiveness of the control algorithm, the frequency response
functions for different controller gain values were obtained exciting the machine with an
impact hammer.

Figure 14 Obtaining frequency response function (FRF)

Figure 15 FRFs for different controller gains

352 I. Garitaonandia et al.

For these tests, one point of the grinding wheel head was selected as impact point. This
point was considered as representative of the zone where the normal cutting forces act.
The excitation was carried out in the infeed direction of the machine. The hammer used
was piezoelectric type (Dytran Instruments Model 5803A). The response was measured
using a piezoelectric accelerometer (Endevco Model 5221) located on the regulating
wheel head. The excitation and response points are shown in Figure 14. FRFs obtained
with gain zero and optimal gain are shown in Figure 15.
In Figure 15, it is possible to observe that the control action produced an appreciable
decrease in peak amplitudes of the different modes. Particularly, the main mode of
chatter (third peak in Figure 15) is reduced considerably. The peak corresponding to the
secondary chatter mode has been omitted in Figure 15 because it remained unaltered,
showing that the actuators have no control authority over this mode.
To demonstrate the ability of the designed control system to reduce the vibration level
of the machine, some cutting tests were performed in chatter conditions and the
accelerations produced in different parts were measured using accelerometers. Figure 16
shows the accelerations produced in the regulating wheel head, both before and after the
application of the control scheme.
The reduction in the vibration level is evident once the control has been applied.
Furthermore, comparing Figure 12 and Figure 16, one can conclude that the theoretical
estimations are in agreement with the experimental results.

Figure 16 Accelerations measured in the regulating wheel head

It should be noted that some limitations were found in the practical implementation:
• The actuators used were saturated for controller gain values lower than the optimal
gain value for the main chatter mode.
• The designed control law had a great influence on the suspension mode shape, which
required a great control force for its active damping. This is not optimal since to
influence in the machine dynamic flexibility, acting in the main mode of chatter is
enough.
 Design of an active control of vibration in a centreless grinding machine 353

7 Conclusions

In this paper, the numerical modelling and experimental procedure to develop an active
control system of vibrations using piezoelectric actuators has been presented.
The theoretical design has consisted of the combined use of FE modelling tools and
state space simulation techniques. The resulting models have predicted, both qualitatively
and quantitatively, the real behaviour of the machine once the actuators have been
incorporated, so the effectiveness of the simulation procedure is demonstrated.
Referring to the practical results, the potential benefit of the application of active
vibration control systems in a centreless grinding machine has been demonstrated,
obtaining a substantial reduction in the vibration level once the control law has been
activated. These results have also shown the practical limitations of the implemented
system, so the theoretical models developed in this paper are an essential requirement to
optimise the design of the control system without the need of costly and time consuming
practical implementations.

Acknowledgements

The authors are grateful for the support of the University of the Basque Country
(EHU08/44-2008).

References
Albizuri, J., Fernandes, M.H., Garitaonandia, I., Sabalza, X., Uribe-Etxeberria, R. and
Hernández, J.M. (2007) ‘An active system of reduction of vibrations in a centerless grinding
machine using piezoelectric actuators’, International Journal of Machine Tools and
Manufacture, Vol. 47, No. 10, pp.1607–1614.
Altintas, Y. and Weck, M. (2004) ‘Chatter stability of metal cutting and grinding’, Annals of the
CIRP, Vol. 53, No. 2, pp.619–642.
Bennett, J.G., Goldman, P., Williams, D.C. and Farrar, C.R. (1994) ‘A comparison of the dynamic
stiffness of the Goldcrown GC-500 grinding machine for three slide designs’, Los Alamos
National Laboratory Report, Los Alamos, New Mexico, USA, LA-12705-MS.
Comstock, T.R., Tse, F.S. and Lemon, J.R. (1969) ‘Application of controlled mechanical
impedance for reducing machine tool vibrations’, Transactions of the ASME, Journal of
Engineering for Industry, Vol. 91, pp.1057–1062.
Ehmann, C. and Nordmann, R. (2003) ‘Comparison of control strategies for active vibration control
of flexible structures’, Archives of Control Science, Vol. 13, Part 3, pp.303–312.
Ehmann, C., Schönhoff, U. and Nordmann, R. (2002) ‘Robust controller synthesis vs. integral force
feedback with collocation for active damping of flexible structures’, Proceedings of ISMA
2002 International Conference on Noise and Vibration Engineering, Leuven, Bélgica.
Fernandes, M.H., Garitaonandia, I., Albizuri, J., Hernández, J.M. and Barrenetxea, D. (2009)
‘Simulation of an active vibration control system in a centerles grinding machine using
reduced updated FE model’, International Journal of Machine Tools and Manufacture,
Vol. 49, Nos. 3–4, pp.239–245.
Furukawa, Y., Shiozaki, S. and Shiga, N. (1973) ‘Suppression mechanism of build-up chatter
vibration – adaptive control of self-excited chatter vibration in centerless grinding
(1st report)’, Journal of the Japan Society of Precision Engineering, Vol. 39, No. 459,
pp.448–454.
354 I. Garitaonandia et al.

Gao, Y., Zhang, D. and Yu, C.W. (2001) ‘Dynamic modeling of a novel workpiece table for active
surface grinding’, International Journal of Machine Tools and Manufacture, Vol. 41, No. 4,
pp.609–624.
Garitaonandia, I., Fernandes, M.H. and Albizuri, J. (2008) ‘Dynamic model of a centerless grinding
machine based on an updated FE model’, International Journal of Machine Tools and
Manufacture, Vol. 48, Nos. 7–8, pp.832–840.
Gosebruch, H. (1987) ‘Active damping of dynamic interactions in grinding machines’, Proceedings
of ASME Design Technology Conference on Vibration Control and Active Vibration
Suppression, pp.147–152, Boston.
Hashimoto, F. and Lahoti, G.D. (2004) ‘Optimization of set-up conditions for stability of the
centerless grinding process’, Annals of the CIRP, Vol. 53, No. 1, pp.271–274.
Hashimoto, F., Zhou, S.S., Lahoti, G.D. and Miyashita, M., (2000) ‘Stability diagram for chatter
free centerless grinding and its application in machine development’, Annals of the CIRP,
Vol. 49, No. 1, pp.225–230.
Hatch, M.R. (2001) Vibration Simulation using Matlab and Ansys, Chapman & Hall/CRC Press,
Boca Raton, Florida, USA.
Mitchell, E. and Harrison, E. (1977) ‘Design of a hardware observer for active machine tool
control’, Transactions of the ASME Journal of Dynamic Systems, Measurement, and Control,
Vol. 99, pp.227–232.
Miyashita, M. (1972) ‘Unstable vibration analysis of centreless grinding system and remedies for
its stabilization’, Annals of the CIRP, Vol. 21, No. 1, pp.103–104.
Miyashita, M., Hashimoto, F. and Kanai, A. (1982) ‘Diagram for selecting chatter free conditions
of centerless grinding’, Annals of the CIRP, Vol. 31, No. 1, pp.221–223.
Nachtigal, C.L. and Cook, N.H. (1970) ‘Active control of machine-tool chatter’, Transactions of
the ASME, Journal of Basic Engineering, Vol. 92, No. 2, pp.238–244.
Preumont, A. (2002) Vibration Control of Active Structures, An Introduction, 2nd ed., Kluwer
Academic Publishers, Dordrecht, Netherland.
Rowe, W.B., Spraggett, S. and Gill, R. (1987) ‘Improvements in centerless grinding machine
design’, Annals of the CIRP, Vol. 36, No. 1, pp.207–210.
Yoshioka, J., Hashimoto, F., Miyashita, M., Kanai, A., Abo, T. and Daito, M. (1985) ‘Ultra
precision grinding technology for brittle materials: application to surface and centerless
grinding processes’, in de Komanduri, R. and Maas, D. (Eds.): Milton C. Shaw Grinding
Symposium, ASME, pp.209–227.
Zhou, S.S. (1994) ‘A dynamic study of the plunge centerless grinding process and its effect on
workpiece out-of-roundness’, PhD Thesis, University of Connecticut.
Zhou, S.S., Gartner, J.R. and Howes, T.D. (1996) ‘On the relationship between setup parameters
and lobing behavior in centerless grinding’, Annals of the CIRP, Vol. 45, No. 1, pp.341–346.
 Design of an active control of vibration in a centreless grinding machine 355

Nomenclature
Vpzt, Vpzt-left, Vpzt_right Applied voltage
ΔL Axial deformation of an actuator
Fpzt Actuator force
Kpzt Actuator stiffness
ga Actuator gain
n Number of actuator piezoelectric discs
d33 piezoelectric constant
g Controller gain
H(s) Controller transfer function
s Laplace operator
ωf Frequency of the filter
ξf Damping ratio of the filter
[Φm] Modal matrix
[Ωm] Frequency matrix
[ξm] Modal damping matrix
F, Fleft, Fright Proportional force to the applied voltage
ym(t) Machine deformation
a(t) Acceleration
{x(t)}, {x(t)
 } State vectors
[Ar], [Br], [Cr], [Dr] Reduced state space matrices