Professional Documents
Culture Documents
net/publication/263779552
CITATIONS READS
0 627
5 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by André Bucht on 10 March 2015.
Keywords:
Mechatronics, Adaptive Spindle, Actuator-Sensor-Unit, Thermal Compensation,
Abstract
This paper presents mechatronic options to increase the productivity and accuracy
in Production. Additional piezo-based components in drive trains can significantly
improve the dynamic behaviour of machine tools. In this article we present a piezo-
based actuator-sensor-unit that is able to reduce uniaxial vibrations in ball screw
driven feed axis of machine tools. Using this component yielded in an increase of
the drives controller parameters by more than 100%. Beyond this piezo-based-
components also possess to increase the accuracy during milling operations and
even allow additional fine positioning movements during machining operations. In
this paper we present an adaptive spindle as an example to benefit both. This is
achieved by an overlaid piezo-based hexapod-kinematics that is mounted between
the machine structure and the motor spindle. Shape Memory Alloys and Phase
Change Materials represent another group of active materials that are usable to
increase accuracy and dynamics. Concerning thermal machine behaviour they are
suitable to compensate thermal deformations. Compared to piezo-based solutions
they rather don’t need additional energy because the activation energy is supplied
by the process.
(a) Scheme of the ASU as part of a feed axis (b) Design concept of the actuator-sensor-unit
Figure 1: System design of the ASU
Using the Laplace transformation allows transforming this equation into the
frequency domain and leads to the commonly used transfer function:
X (s ) 1 1
Gs (s ) = = ⋅ 2 2 . (2)
Fi (s ) c T s + 2 DTs + 1
The parameter T represents the reciprocal of the resonance frequency and D
typifies the damping. A direct velocity feedback as shown in Figure 1 (a) results in
the following closed loop transfer function:
Gs (s ) 1 1 (3)
Gwx (s ) = = ⋅
1 + Gs (s ) ⋅ k DVF ⋅ s c k
T 2 s 2 + 2 DT + DVF s + 1
c
The damping term of the equation now consists of the sum system damping 2DT
and an additional damping part kDVF/c, that can be inserted by an appropriate
actuator. Every other parameter of the system remains untuned. The advantages of
this approach are:
• No communication required with the overlaid drive control.
• It is not necessary to model the system due to easy tuning of the controller
• Stability can be guaranteed in conditions at which sensor and actuator create
a co locative pair.[22]
The easiest and the most convenient way to control the ASU is by measuring the
acceleration of the feed axis’ table. Since it is necessary to integrate the
acceleration signal to get the velocity this approach leads to an I-controller.
However, constant components in the signal would force the controller in its
nd
limitation. To avoid these effects, the controller must be enlarged by a 2 order
bandpass filter. The resulting transfer function is:
1 2 D / ω0 ⋅ s . (4)
GIBP (s ) = ⋅
TI s 1 + 2 D / ω0 ⋅ s + 1 / ω02 ⋅ s 2
The centre frequency is initially set to the eigenfrequency of the system and the
bandwith should cover the frequency range of the resonance. The amplification can
be set by decreasing TI [17]. The direct velocity feedback (active damping) is only
suitable to damp the resonance frequency. In case of static deformations these can
be compensated by an algorithm called positive position feedback (active stiffness).
This can be achieved by a double integration of the acceleration signal. As shown in
[17] it is possible to combine the two approaches.
(a) One-mass oscillator and corresponding control loop (b) Impact on the frequency response
Figure 3: Pole placement as control concept
The two approaches where implemented in the model. The controller considers also
the voltage limitations of the amplifiers and the sampling time. Figure 3 (b) shows
the simulation results of the feed axis model.
It is obviously visible that the active damping approach reduces resonance to
almost zero. The combination of the two approaches does not show a significant
difference in the input frequency response, but a significant decrease of flexibility in
the disturbance frequency response at low frequencies.
60
Standard Milling
Figure 6: CAD model of the adaptive spindle support (left) and real component (right) mounted in
a Parallel-Kinematic Machine (PKM).
40
Gφx,φx calculated
30
20
10
0
-10
-20 1 2
10 10
Frequency [Hz]
Figure 8: Comparison of the FRF resulting from a reference displacement in ϕx (tilting around x-
axis).
Figure 8 shows the comparison between measured and calculated data. The
Frequency Response Function (FRF) was identified by tilting the spindle around the
x-axis. The first eigenfrequencies in the model are close to the measured ones.
Therefore, it can be assumed that the calculated also meet the real eigenforms. For
the control design especially the ones that affect more than one direction are
significant. The eigenform at 204Hz for example Figure 9 has a main deflection in
ϕx-direction, but also affects movements in ϕy and y-direction. Such couplings have
to be considered in the control approach.
To design a decoupling network it is necessary to describe the transfer behaviour of
an elastic structure in the frequency domain. These transfer functions can be written
in modal parameters as sum of the eigenmodes as shown in
n Φ ir Φ jr 1
Gi , j ( s ) = ∑ ⋅ 2 2
(5)
r =1 M r s + 2 Λ r ω r s + ωr
The parameter Φ represents the components of the eigenvector. Λr typifies the
modal damping, ωr the eigenfrequency and Mr the modal mass of the r-th
eigenmode. Gi,j was examined by measuring the transfer functions from all inputs to
outputs. The modal parameters can be extracted from the measured results by
using appropriate software. In this way it is possible to get a numeric model which is
based on the physical dependencies of the structure and to allocate the measured
transfer functions to the calculated eigenforms.
Figure 9: FE-Model of the spindle and eigenform of the 5th eigenmode at 204 Hz. (tilting x-axis).
The matrix G contains the transfer functions Gi,o from input i to output o and has the
dimension 5x5. To examine the couplings between the entire axes, measurements
of all transfer functions have been performed. Regarding the normalized static
transfer values clarify that there are two main couplings. First, a drive signal on the
y-axis causes an unwanted output in ϕx-direction and second, an input signal in x-
direction results in a displacement in ϕy-direction. Considering this, the complete
system can be described by the matrix given in Equation (7) or by the block
diagram shown in Figure 10 right.
Gx , x ( s ) 0 0 0 0
(7)
0 G y, y ( s ) 0 0 0
G (s) = 0 0 Gz , z ( s ) 0 0
0 G y ,ϕ x ( s ) 0 Gϕ x ,ϕ x ( s ) 0
G ( s ) 0 0 0 Gϕ y ,ϕ y ( s )
x ,ϕ y
GD(s) is a diagonal matrix with the transfer functions of Gi,i as elements. The
conditions to reach the aspired behaviour can be calculated by a coefficient
comparison given by
GD (s) = G (s) ⋅ D(s) (10)
The known matrices are GD and G and the aspired decoupling network is D which
can be calculated by transposing equation 10 to
−1
D(s) = G (s) ⋅GD (s) (11)
The inverse transfer function matrix G-1 exists if the determinant is not zero, which
is usually the case.
Figure 10: Block diagram of the plant with added decoupling network.
(a) Measured FRF for tilting around y-axis (b) structure of the feedback loop
Figure 11 Decoupling results.
With an implemented decoupling network it is possible to develop independent
position controllers for every single Cartesian axis. A suitable approach for position
control is the PID-controller. This approach is not model-based that makes it very
robust against time variant differences in the transfer function. The control design
process is discussed in [26]. The five PID-controllers can be concentrated in the
controller matrix C. The simplified control scheme is shown in Figure 11.
Figure 12 shows the result for a spindle speed of 3.900 rpm. The measured cutting
force was about 20N. The reachable maximum force was not determined yet. An
operation frequency above 130Hz causes problems in the control loop due to the
close resonance frequencies at 171Hz and above. The phase shift in the control
loop increases over the critical value of 180 degrees and the control loop gets
unstable. Lowering the proportional gain in the controller avoids that but also results
in a lower bandwidth. The actual maximum operation frequency in tilting mode is
therefore limited in this concept to approximately 130Hz.
Thermal SMAs offer the special ability to “remember“ and re-assume their original
shape following permanent plastic distortion below a specific critical temperature by
means of heating up above this temperature. A reversible austenite-martensite
phase transformation is required for the development of the shape memory effect.
Analogous to steel the high temperature phase of the material is also described as
austenite and low temperature phase α as martensite. In an ideal situation the
austenite β phase is converted into the martensite α phase as a result of shear. Due
to diffusion-free rearrangement processes in relation to the atoms this generates a
change in the stacking sequence of the crystal lattice levels and therefore enabling
a change in the structure of the crystal lattice. Consequently, two different stress-
strain-curves exist as shown in Figure 14. In the low-temperature phase a small
Hook region is followed by a so-called plateau-stress. in this case, the material can
be easily deflected almost without increasing the applied external stress. After
setting the stress to zero, a plastic deflection remains to the material. Heating the
material causes the described phase transformation and results in a completely
different stress-strain-behaviour. During the phase transformation from martensite
to austenite (heating) the material is able to perform mechanical work (see Figure
14(b)). The amount of work depends on the mechanical boundary conditions. The
normal operation mode is to use the material in an arrangement with a spring, as it
is shown in Figure Figure 14(a) or a mass. The mechanical design of such
actuators has been described previously and details can be found in [2].
Setting the pre tension in common ball screw drives lead to a conflict of aims
between high pre tension to achieve precise positioning behaviour and low pre
tension to achieve low friction forces. The durability of ball-screw-drives is
characterized by abrasion, material fatigue and basically the applied load [7].
Operating a ball-screw-drive with high pre tension causes an increase of friction and
Figure 16 (b) shows the temperature profile of such system during positioning
operations. That causes negative effects regarding the durability and accuracy due
to an increasing backlash. This problem can be solved by varying the pre tension
during operation. Currently known approaches range from conventional systems
using servo drives [6] to systems using piezoceramic actuators [24]. These
solutions all together necessitate implemented actuators that are considering the
efforts quite honestly uneconomic.
According to the approaches discussed in chapters 3.1.3 and 3.1.4 there are two
different system designs. The fully controllable approach, shown in Figure 16 (a)
uses heat foils for heating the actuator and a force sensor to measure the pre
tension. With a suitable controller it is possible to follow a defined value of the pre
tension. Due to the needed actuator mass, the bandwidth of the control is strictly
limited. Nevertheless it is possible to react on changes in the environmental
conditions.
The self-adapting adapting approach, shown in Figure 16 (b), does rather not use
any external elements. The control loop is only closed by the structure of the ball-
screw drive, the SMA-actuator and the friction heat. As it was mentioned in chapter
3.1.4 the system design rather has to be focused on the thermo-mechanical
behaviour of these elements than a control algorithm. Even though an active control
of the pre tension is not possible, the approach ensures a constant pre tension
independent of environmental changes and temperature changes of the ball screw
drive. Both approaches are currently under investigation.
3.2 Phase-Change-Materials
Phase change materials (PCM) are substances with a high heat of fusion which,
melting and solidifying at a certain temperature, is capable of storing and releasing
large amounts of energy. Heat is absorbed or released when the material changes
from solid to liquid and vice versa; thus, PCMs are classified as latent heat storage
units. PCMs latent heat storage can be achieved through solid-solid, solid-liquid,
solid-gas and liquid-gas phase change. However, the only phase change used for
PCMs is the solid-liquid change.Initially, the solid-liquid PCMs behave like sensible
heat storage materials (SHS), their temperature rises as they absorb heat. Unlike
conventional SHS, however, when PCMs reach the temperature at which they
change phase they absorb large amounts of heat at an almost constant
temperature. The PCM continues to absorb heat without a significant raise in
temperature until all the material is transformed to the liquid phase. When the
ambient temperature around a liquid material falls, the PCM solidifies, releasing its
stored latent heat. A large number of PCMs are available in any required
o
temperature range from -5 up to 190 C. Within the relevant range of 20° to 50°C,
some PCMs are very effective. They store 5 to 14 times more heat per unit volume
than conventional storage materials such as water [9].
(a) measurement setup (b) measurement results
Figure 17: PCM-filled ball screw drive
To investigate the potential of PCM for using in machine tools a conventional ball
screw drive was modified to carry an amount of PCM. Around of the flange nut an
additional housing was mounted that a thin cavity exists between the flange nut and
the housing. The cavity is filled with a commercial available PCM as it is shown in
Figure 17 (a). The mass of the PCM was 40g. The housing was made of aluminium
and has a weight of 125g. The flange nut consists of steel with a weight of 730g.
To measure the influence of the PCM the temperature of the flange nut during a
typical machining cycle was measured using thermo-elements arranged at the
flange nut and the housing. The measurement results are shown in Figure 17 (b).
The ball screw drive without the PCM shows the expected behaviour. After a fast
rise in the first minute of the machining cycle the temperature slope decreases. The
flange nut reaches a temperature of 40.5°C at the end of the machining cycle. The
measurement with the PCM-filled housing shows a similar behaviour up to 17 min
of the measurement time. After reaching a temperature of 37°C de PCM starts to
transform from solid to liquid. Until the temperature reaches 38°C the PCM works
as a heat sink that limits the temperature of the flange nut. After 30 min the PCM is
not able to store more heat energy. So the PCM behaves like a common material
and the temperature of the flange rises again. Until the end of the machining cycle,
the temperature of the PCM filled flange nut remains 1.5°K lower than the common
nut. The behaviour in the cool-down cycle shows even more differences. The PCM-
filled ball screw drive has a significantly slower temperature decrease. So it is
possible to avoid temporary temperature peaks, the temperature remains in a
defined band. The potential of this approach will be investigated in the future.
4 Summary and Outlook
Mechatronic approaches possess a supreme development methodology for
machine tools. The future developments considering the basic structure –
mechanics/ materials, transformation systems – sensor/actuator systems and data
processing will characterise the development of “intelligent” machine tools. Within
the next few years, the significance of the emerging trends like self-optimising
adaptronic components or more efficient control systems for model supported
compensation of machine errors and process control will increase multi-fold.
In the past, huge efforts were made in developing piezo-based components to
increase the accuracy and productivity of machine tools. As a result there are
numerous application solution available as for instance, the actuator-sensor-unit or
the adaptive spindle support. However, the applications for commercial available
machines are sparse. The main reasons therefore are the harsh industrial
environment and the higher costs of piezo components. The reliability of machine
tools is one of the main issues for machine manufacturers. Despite thorough
investigations, the reliability of piezo components is not sufficient. Although the
number of suppliers which offer commercial actuators and amplifiers increased in
the last years, the costs of these components are still the essential cost drivers of
adaptronic systems. The resulting system costs are not tolerated by the machine
manufacturers. The prospective work has to deal with these aspects in order to
transfer piezo applications from the laboratory to industrial applications.
The approaches for thermal compensations using thermo sensitive materials like
SMA’s or PCM are rather unexplored. The main research is therefore currently
focused on discovering the interactions between the active material and the thermo-
mechanical structure. It is a mighty challenge because the level of integration is
even higher than it is for the piezo systems. Especially the aspired self adapting
systems which use the process heat are extremely complex. The absence of any
external energy source and control component decreases the system costs but
increases the demands on engineering complexity. The challenge is to exploit the
full capabilities already in the early design phase. The design tools for these
adaptronic systems cannot be reduced to the sum of modelling, simulation of
individual domains and securing appropriate exchange of data. A closed work chain
has rather to be developed. Therefore a complete exploration of the thermo-
mechanical interactions as well as the material behaviour and the aspects of
designing a structure-inherent control loop is necessary. Achieving this scientific
level is the main issue of the prospective work.
Literature
[1] Bachmann electronic GmbH Dokumentation Servosystem S1., Feldkirch(A)
[2] Besselink, P.A. (1996) Procedure for the Calculation of the Geometry of a
Resistance Heated NiTi-Actuator. Proceedings of the 5th International
Conference on New Actuators.
[3] Brecher C (2002) Vergleichende Analyse von Vorschubantrieben für
Werkzeugmaschinen. Dissertation RWTH Aachen.
[4] Denkena, B.; Möhring, H.-C.; Will, J. C. (2005) Design and layout of an
adaptronic spindle system. CIRP Journal of Manufacturing Systems,
34/3:247-257.
[5] Drossel, W.-G.; Wittstock, V. (2008) Adaptive Spindle Support for Improving
Machining Operations. 58th General Assembly CIRP 2008, Manchester (UK),
CIRP Annals 57 Nr.1, S.395-398
[6] Frey, S.; Walther, M.; Verl, A. (2010) Periodic variation of preloading in ball
screws. Production Engineering.
[7] Golz, H.-U. (1990) Analyse, Modellbildung und Optimierung des
Betriebsverhaltens von Kugelgewindetrieben. Dissertation. Karlsruhe: WBK-
Verlag (Forschungsberichte aus dem Institut für Werkzeugmaschinen und
Betriebstechnik der Universität Karlsruhe, 31).
[8] Groß H, Hamann J, Wiegärtner G (2000) Elektrische Vorschubantriebe in der
Automatisierungstechnik. Publicis MCD Verlag.
[9 Hiebler S.(2007) Kalorimetrische Methoden zur Bestimmung der Enthalpie
von Latentwärmespeichermaterialien während des Phasenübergangs.
Dissertation Technische Universität München, Bayerisches Zentrum für
Angewandte Energieforschung e.V. Abteilung 1: „Technik für Energiesysteme
und Erneuerbare Energien“.
[10] Janocha H (1999) Adaptronics. Springer-Verlag, Berlin.
[11] Jendritza D (1998) Technischer Einsatz Neuer Aktoren. Expert-Verlag,
Rennigen-Malmsheim.
[12] Kranz, B., (2009) Zustandsraumbeschreibung von piezo-mechanischen
Systemen auf Grundlage einer Finite-Elemente-Diskretisierung. ANSYS
Conference & 27th CADFEM Users’ Meeting, November 18-20, 2009
Congress Center Leipzig, Germany, pp 2.9.16
[13] Lenk A, Pfeifer G (2001) Elektromechanische Systeme. Springer-Verlag,
Berlin.
[14] Namazi, M. , Altintas, Y., Abe, T., Rajapakse, N.(2007) Modelling and
identification of tool holder-spindle interface dynamics. J. o. Machine Tools &
Manufacture, 47/9:1333-1341.
[15] Neugebauer R, Wittstock V, Illgen A, Naumann G (2006) Piezo-based
Actuator-Sensor-Units for Uniaxial Vibration Reduction in Machine Tools.
Production Engineering 13: 211-216
[16] Neugebauer, R.; Denkena, B.; Wegener, K. (2007) Mechatronic Systems for
Machine Tools, Annals of the CIRP, 56/2.
[17] Neugebauer R, Bucht A, Illgen A, Wittstock V (2007) Schwingungsreduktion
mit Aktor-Sensor-Einheiten. wt Werkstattechnik online 11/12: 901-907.
[18] Neugebauer R., Pagel K., Bucht A., Wittstock V., Pappe A.(2010) Control
Concept for Piezo-Based Actuator-Sensor-Units for Uniaxial Vibration
Damping in Machine Tools. Production Engineering XX
[19] Neugebauer, R.; Drossel, W.-G.; Bucht, A.; Kranz, B.; Pagel, K.: Control
design and experimental validation of an adaptive spindle support for
enhanced cutting processes, 60th General Assembly CIRP 2010, Pisa (It),
CIRP Annals 59 (in Press)
[20] Neugebauer, R., Bucht, A., Pagel, K., Jung, J.,: Numerical simulation of the
activation behavior of thermal shape memory alloys, Industrial and
commercial applications of Smart Structures Technologies, Proc. of SPIE,
San Diego, CA, ( 2010)
[21] O'Neal, G.P.; Min, B-K.; Pasek, Z.J.; Koren, Y.(2001) Integrated
Structural/Control, Design of Micro-Positioner for Boring Bar Tool Insert. J. o.
Intelligent Material Systems and Structures, 12:617-627.
[22] Preumont A (1997) Vibration Control of Active Structures. Kluwer Academic
Publishers.
[23] Schulze, M: (2008) Elektrische Servoantriebe. Carl Hanser Verlag München.
[24] Weule, H.; Golz, H.-U. (1991: Preload-Control in Ball Screws - A New
Approach for Machine Tool Building. CIRP Annals - Manufacturing
Technology, Jg. 40/1, H. 1, S. 383–386.
[25] Wittstock V (2006) Entwicklung piezobasierter Komponenten für den
Maschinenbau. Konstruktion 3: 61-66.
[26] Wittstock V (2007) Piezobasierte Aktor-Sensor-Einheiten zur uniaxialen
Schwingungskompensation in Antriebssträngen von Werkzeugmaschinen.
Verlag Wissenschftliche Scripten, Zwickau.