Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

CIRP Annals - Manufacturing Technology 59 (2010) 395398

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

Development of an active clamping system for noise and vibration reduction


J. Hesselbach (2), H.-W. Hoffmeister, B.-C. Schuller *, K. Loeis
Institute of Machine Tools and Production Technology (IWF), Technische Universitat Braunschweig, Langer Kamp 19b, 38106 Braunschweig, Germany

A R T I C L E I N F O

A B S T R A C T

Keywords:
Machine tool
Vibration
Adaptronic

Noise emissions of up to 110 dB(A) occur during the machining of composite boards. Vacuum clamping
systems are predominantly used for machining particle boards on woodworking machining centers.
These clamping systems enable a good accessibility to the workpiece edges during milling. As a result to
this clamping method the boards have non-clamped areas. Consequently they vibrate over a wide
frequency domain during machining. The quality of the particle board edges is reduced and a high noise
level is induced by these vibrations. An active clamping system based on piezo-stack actuators has been
designed and implemented to reduce these vibration amplitudes. All required steps of its development,
the localization of the vibrations and the noise emission, the mechanical and control design and the
system integration into the machine table, are presented in this paper. The achieved results of this
research demonstrate the signicance of active systems in machine tools.
2010 CIRP.

1. Introduction
Current developments and research in the eld of stationary
woodworking machines mainly focus on an increase in productivity. These developments are always associated with an increase
in machine performance, which is connected with setting up
higher feed velocities. These machines actually reach rapid feed
velocities of up to y f 160 m=min which cannot always be
transferred to working feed velocities. To maintain or to get a
better workpiece quality at these velocities, higher cutting speeds
are required as well.
Higher velocities and accelerations of each machine axis induce
higher dynamic loads, which affect the entire machine system, the
environment and the peripherals. These loads generally enhance
structural vibration amplitudes which affect the production quality
negatively and induce a high noise emission during machining.
The minimum requirements for the protection of workers from
risks to their health and safety arising out of exposure to vibrations
and noise were tightened with the ratication of the Directive 2003/
10/EC by the European Parliament. More precisely, the limit and
action values in respect to the daily noise exposure levels and peak
sound pressure levels were reduced by 5 dB(A) [1]. Besides the
machining quality, the sound radiation of machines becomes an
important assessment benchmark for machine tools. A recent
analysis revealed that many workplaces in the woodworking
industry exceed these new legal limits by far. Technological
advancements have to be made to improve the performance and
simultaneously reduce the vibration and noise emission of these
machines with regard to these facts. Local encapsulation or passive
damping methods generally used to counteract this problem do not
reduce the noise level as desired. Besides, such solutions particularly
for woodworking machining centers are limited to design aspects.
An improved performance can only be realized by new machine

* Corresponding author.
0007-8506/$ see front matter 2010 CIRP.
doi:10.1016/j.cirp.2010.03.079

component design and new system developments. These may be


achieved by the integration of active or adaptronic systems into the
mechanical components of machine tools. These systems, which are
also known as smart structures, use multi-functional materials that
can be used as sensors or actuators. By means of e.g. piezoelectric
properties, they can adapt independently to changing inuences of
the surrounding or operation conditions [2,3]. In opposition to
passive vibration reduction systems, the latter ones are able to
control the system behavior during machining by an adequate
material and structure integration. Therefore, they are able to
interrupt the transmission path of the vibration energy.
Based on a systematic vibration and noise generation analysis
the design and test of an adaptronic system at wood machining
centers were investigated.
2. Vibrations on woodworking machining centers
The structure of woodworking machining centers is predominantly designed as an arm structure. This kind of structure is ideal
for a complete machining of different workpiece geometries. The
workpiece materials which are machined on these centers are
mostly derived timber products for furniture industry and timberframe constructions, e.g. particle boards and MDF (medium
density berboard) panels. Stationary woodworking machining
centers predominantly use a vacuum system as clamping device.
By using this system, workpieces are neither damaged nor
deformed during milling, drilling or sawing processes. The system
allows an arbitrary positioning of each suction block and a good
accessibility to the workpiece edges.
Milling processes involve two kinds of motions. The primary
motion is realized by the revolution of the end milling tool and the
secondary is the infeed motion. The necessary cutting force for chip
removal has to be absorbed by the tool cutting edge and the
workpiece. The cutting force is composed of an active and a passive
force, which is perpendicular to the working plane. The active force
Fa can be divided into two components Ff and Fc which point into

396

J. Hesselbach et al. / CIRP Annals - Manufacturing Technology 59 (2010) 395398

Fig. 1. Cutting force components and vibration propagation during milling.

the direction of the feed velocity y f and in the direction of the


cutting speed yc (Fig. 1).
Therefore, the direction of the active force changes with the
entrance angle of the cutting edge. Due to discontinuous cutting
(with number of teeth z = 1), high dynamic loads also occur in the
tooth contact zone. These loads which are induced by the cutting
forces lead to thermal loads and mechanical dynamic stresses and
strains. They spread out over the cutting tool and workpiece side
into the machine structure (Fig. 1). Depending on the material
properties of each component these loads consequently induce
forced vibrations into the overall system composed of the machine
structure, the tool and the workpiece. The dynamic behavior of the
system is described by mass, stiffness and damping characteristics.
Each structure depending on the size of these values yields
particular mode shapes at specic natural frequencies independent of the dynamic loads which are acting on the structure.
The natural frequencies and the mode shapes of the machine
structure and the clamped workpiece were determined by an
experimental modal analysis. Analysis results show that the
workpiece has a 100 times higher dynamic compliance compared
to the tool and the machine structure, especially with this
clamping conguration. However, the main disadvantages of the
above described clamping method are the free and overhanging
areas of the boards. These non-clamped areas raise the dynamic
compliance of the workpieces. Consequently they are excited
during high rate processing over a wide frequency range, which
increase the risk of chipping along the workpiece edges.
The frequencies that occur during machining were analyzed.
These have a signicant inuence on the dynamic behavior of the
workpiece and consequently on the vibrations and noise emission.
During the milling process, the vibration amplitudes were

Fig. 2. Vibration amplitude (spectrogram) at the workpiece surface.

measured with accelerometers on the workpiece surface. The


rotation speed of the tool was na = 18,000 rpm. The spectrogram in
Fig. 2 of a measured signal shows the dominant vibration
frequencies during idling (run-up and run-out) and machining.
The maximum vibration amplitude is as expected at the
fundamental frequency of 300 Hz. In this case a forced vibration
propagates all over the workpiece due to its higher compliance.
Besides the fundamental frequency of 300 Hz Fig. 2 shows that in
the frequency range of 05 kHz several harmonics are present.
Therefore, the vibration is identied as forced impulse vibration.
Due to this kind of excitation, the workpiece will oscillate
during machining with a complex mode shape. The mode shape is
a linear combination of all the modes at each natural frequency
which lies in the vicinity of the fundamental excitation frequency
and their harmonics.
3. Noise emission on woodworking machining centers
The process forces which occur during machining induce direct
and indirect airborne sound in the tooth contact zone.
The direct airborne sound is composed of the idling sound of the
tool and the impulse sound which results from the contact
between tool and workpiece. The second type of sound is
generated from the material displacement and chip removal
process during machining. Both sounds are emitted directly to the
air space of the machine environment (Fig. 3).
The indirect airborne sound is induced from the impulse
changing forces, which occur during machining. These forces
induce vibrations (waves) in a certain frequency range in the tool
and in the workpiece. On the one hand, a part of the mechanical
waves is transmitted by the tool into the machine structure. Hence,
this part is insignicant due to the higher stiffness of the tool and
machine structure in relation to the stiffness of the workpiece. On
the other hand, the changing forces induce a transversal motion
and a rotational motion of each material particle of the workpiece.
Due to the applied clamping method, the properties and the
geometry of the workpiece, these vibrations propagate as bending
waves over the entire workpiece. Thereby, the propagation
velocity of the bending waves cB is a function of the bending
stiffness B0 per unit area, the site-related mass m00 of the workpiece
and the excitation frequency:
s
p 4 B0
cB v
(1)
m00
The structure-borne sound resulting from these waves is
transmitted through the workpiece material and radiated through

J. Hesselbach et al. / CIRP Annals - Manufacturing Technology 59 (2010) 395398

397

Fig. 5. Multi-body model of the adaptronic clamping system.

Fig. 3. Direct and indirect airborne sound.

the workpiece surface into the air space of the machine


environment (Fig. 3). To nd out if the dominant acoustic source
is located in the area of the tooth contact or in the area of the
workpiece surface, several measurements and analyses were done
with an acoustic camera, a measurement system which uses
beamforming as signal processing technique to visually localize
acoustic emission. The image sequences in Fig. 4 show the location
of the maximum sound radiation during idling and material
removal.
During idle running, the acoustic emission is generated mostly
from the tool rotation. The location of the dominant acoustic
emission changes instantly after the tool gets in contact with the
workpiece. The dominant noise emission moves in the middle of
the workpiece and above. These measurements demonstrate that
the dominant noise emission during milling on woodworking
machining centers is generated mainly from the workpiece and not
only from the tool or tooth contact.
4. Mechanical and control design
To reduce the vibration and the structure-borne sound of the
workpiece, an adaptronic clamping system which is composed of
four sensors, a workpiece clamped on four suction blocks and four
actuators within each of these suction blocks was developed. The
design and behavior of the system were rst modeled and analyzed
in a multi-body-simulation environment. In the simulation model
the workpiece was displayed as a nite element model. The force
acting on the workpiece was modeled as input vector at the
workpiece edges. Likewise, each piezoelectric actuator was
modeled as a spring-damper element and a translational joint
(Fig. 5). The connection between these joints and the exible

Fig. 6. Block diagram of Filtered-X LMS algorithm using RLS cancellation path
modeling.

model of the workpiece was done over additional nite elements


and interface points. The simulation results of the structure
behavior at the sensors reveal that with this actuator conguration
the vibration amplitudes of the workpiece at 300 Hz can be
reduced about 14 times. Based on this result four active suction
blocks were designed. Due to the available design space each
piezoelectric actuator was placed in horizontal position and its
actuating movement is later on translated into a vertical one
(Fig. 7). A multiple-input multiple-output (MIMO) system was
chosen [4] as control technique. Using a Filtered-X LMS (Least
Mean Square) and/or RLS (Recursive Least Square) algorithm the
control system is intended to be adaptive in order to cope with
various machining parameters and workpiece dimensions (Fig. 6).
Each control signal from a single actuator which propagates
through the physical system S(z) is modied by the response
characteristics of the control source and the workpiece between it
and the error sensor. The inuence of all of these can be lumped into
a single secondary path transfer function S0 (z) which is constructed
through a preliminary system identication [4]. As illustrated in
Fig. 6 the error signal e(n) measured by the error sensor is a mixture
of signals from P(z) and S(z). P(z) is the primary path from the
reference signal depending on the rotating frequency of the milling
tool to the error sensor, whereas S(z) is the single secondary path
between each piezoelectric actuator and each error sensor.

Fig. 4. Localization of acoustic emission during idling (left) and machining (right).

J. Hesselbach et al. / CIRP Annals - Manufacturing Technology 59 (2010) 395398

398

Fig. 7. Experimental setup.

Assuming that W(z) is a FIR- (nite impulse response) lter of tapweight length N, the control signal y(n) is computed as
yn wT nxn

(2)

w w0 n; w1 n; :::wN1 nT

(3)

Fig. 8. Vibration reduction during milling process of MDF.

is a tap-weight vector and


xn xn; xn  1; :::xn  N 1T

(4)

is an N-sample reference signal vector. The tap-weights of the


adaptive lter W(z) are updated using the LMS algorithm
wn 1 wn  mx0 nen

(5)

where m is the corresponding step-size parameter and x0 (n) is an Nsample ltered reference signal. The ltered reference signal x0 (n)
is an approximation result after the inuence factors of the
cancellation path described as follows:
0

x0 n S zxn

(6)

where S z is transferred from the RLS cancellation path modeling


S0 (z). Due to the higher stability and less computation time the LMS
algorithm is used in the control process. The identication process
is left to the RLS algorithm. This algorithm identies the secondary
path faster and more precisely. This combination leads to the
Filtered-X LMS controller with the RLS identication, shortened to
the FXLMS-RLS.
5. System integration and experimental results
In order to examine the practical issues associated with the
MIMO-FXLMS-RLS controller, several experiments during machining were carried out. Fig. 7 shows the experimental setup with the
MIMO-FXLMS-RLS structure using four laser triangulation sensors
(LTS) as error sensors.
The actuators and the sensors were located relatively close to
the disturbance source at the milling position, yet the machining
should not be disturbed. The workpiece (MDF, 600 mm  600 mm
 18 mm) was clamped on the active suction blocks. It was milled
at a workpiece edge with a cutting depth of 2 mm and a rotation
speed of the tool of 18,000 rpm.
During the milling process, the vibration amplitude of the
workpiece was dominated at a rotational frequency of the tool of
300 Hz. Therefore, the error signals were ltered by a narrow bandpass lter in the corresponding frequency. Each secondary path from
each actuator to every error sensor was identied at the frequency of
300 Hz as well. Since an improved workpiece quality is the primary
research goal, the vibration amplitudes of the workpiece at the
milling position (LTS 1 and LTS 2) become a priority to be reduced. A
maximum vibration reduction was achieved by 20 dB at LTS 1 and
25 dB at LTS 2 (Fig. 8). At the same time, the vibration amplitude at
sensors LTS 3 and LTS 4 were reduced.
Furthermore, in order to capture the noise radiation during the
machining a microphone was subsequently positioned over the
workpiece surface. The dominant sound pressure levels were
measured likewise at the tools rotation frequency of 300 Hz and

Fig. 9. Waterfall diagram of noise level reduction during milling process.

its ve harmonics at 600 Hz, 900 Hz, 1200 Hz, 1500 Hz and 1800 Hz.
The active clamping system was also investigated with the aim to
reduce the noise levels at these frequencies by employing an
accelerometer in the middle of the workpiece surface as an error
sensor. Actuator 1 and actuator 2 controlled the frequencies of
300 Hz and 600 Hz, whereas actuator 3 and actuator 4 handled the
frequencies of 900 Hz and 1200 Hz, respectively. With this conguration a maximal noise level reduction at 300 Hz by 20 dB(A) and an
overall noise level reduction of 4 dB(A) was achieved (Fig. 9).
6. Conclusions
The vibration and noise analysis presented in this paper prove
that during milling of composite boards on woodworking
machining centers the workpiece is excited with forced impulse
vibrations. These lead to high vibration amplitudes and noise
radiation of the workpiece. With regard to the analyzed results, an
adaptronic clamping system based on piezo-stack actuators was
designed and integrated into the machine table. The experimental
results verify that by the usage of such a system an active noise and
vibration reduction can be successfully realized.
Acknowledgements
The presented work was funded by the German Research
Foundation (DFG), within the Priority Program 1156. The authors
would like to thank for this support.
References
[1] Directive 2003/10/EC. Ofcial Journal of the European Union .
[2] Janocha H (1999) Adaptronics and Smart Structures. Springer, Berlin.
[3] Neugebauer R, Denkena B, Wegener K (2007) Mechatronic Systems for Machine
Tools. Annals of the CIRP 56(2):657686.
[4] Hansen CH, Snyder SD (1997) Active Control of Noise and Vibration. E & FN Spon,
London.

You might also like