See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/263779552

Mechatronic options for increasing productivity and accuracy in production

Conference Paper · January 2010

CITATIONS READS

0 627

5 authors, including:

Welf-Guntram Drossel Kenny Pagel

Fraunhofer Institute for Machine Tools and Forming Technology IWU Fraunhofer Institute for Machine Tools and Forming Technology IWU
339 PUBLICATIONS   1,119 CITATIONS    48 PUBLICATIONS   183 CITATIONS   

SEE PROFILE SEE PROFILE

André Bucht Tom Junker

Fraunhofer Institute for Machine Tools and Forming Technology IWU Fraunhofer Institute for Machine Tools and Forming Technology IWU
61 PUBLICATIONS   221 CITATIONS    15 PUBLICATIONS   30 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

smart³ - materials, solutions, growth View project

fault tolerante control View project

All content following this page was uploaded by André Bucht on 10 March 2015.

The user has requested enhancement of the downloaded file.

Mechatronic Options for Increasing
Productivity and Accuracy in Production
Drossel, W.-G.; Neugebauer, R.; Pagel, K. Bucht, A., Junker, T.
Fraunhofer IWU, Chemnitz, Germany

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.

1 Piezo-based Actuator-Sensor-Units for Uniaxial

Vibration Damping in Machine Tools
Drive trains of machine tools are mechatronic systems that consist of mechanical
power transmissions and position controlled electric drives. Due to the limited
stiffness and the low damping of the mechanical components the dynamics of
electromechanical drives is limited. The characteristics of optimized drive trains are
usually a compromise between lightweight construction to achieve high acceleration
and stiff mechanical components to guarantee a high manufacturing quality.
A typical structural element is the controlled feed axis equipped with ball screw
drive systems. Two major demands have to be met by the structural design. In
order to reduce the process time, spindle-nut-systems have to move the tool
carriage between two machining operations with as high as possible velocity and
acceleration. To meet this requirement, all transmission parts are designed to
possess small masses or moments of inertia. This characteristic, however, is in
conflict to the second demand: high precision movements during machining
operations that require high stiffness. The frequency characteristic, which can be
completely different, depends on the acceleration during positioning operations, the
movable work piece mass, and the cutting or process force, respectively. In spite of
the optimization of mechanical transmission parts and the position control loop, the
axial stiffness of the free length of the spindle causes axial vibrations. The ever
increasing demand on both precision and machine output further intensifies this
conflict. Therefore, a new active component was developed and implemented into
the feed axis, which is placed between the ball screw nut and the tool as seen in
Figure 1(a).

(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

1.1 The Actuator Sensor Unit

Equipped with a differential setup of piezo actuators, the actuator-sensor-unit (ASU)
acts as an additional drive in direction of the main drive. Piezo stack actuators with
their wide frequency range and extraordinary high forces are suitable drives
concerning the expected disturbance level. Figure 1 (a) shows the basic electro-
mechanical principle of the transformation of actuator stroke to movement of the
feed axis. The detailed design process of the component is described in [26],[25].
Figure 1 (b) shows the design of the ASU including a solid state joint, actuators and
sensors.

1.2 Model of the controlled feed axis including ASU

Piezo ceramics are dielectric materials, which generate charges when an external
mechanical force is applied (sensor effect). The application of an electric field on
the other hand leads to a deformation (actuator effect). Both effects rest upon the
atomic dipoles structure and are equally effective [10],[11]. This behaviour is
described with electromechanical state equations. The equations couple the electric
behaviour of a capacitor with the one of an elastic body. The electromechanical
state equations lead to the two port model for discrete actuators. According to the
theory of an electromechanical converter [13], this model can be transformed to a
complete mechanical model. Figure 2 shows the resulting model of the piezo
actuator (white block). Using this phenomenology, the mechanical model of the feed
axis can be extended by active components and the piezo characteristics as
described in [13]. According to the electromechanical scheme of the ASU and the
feed axis as shown in Figure 1 the model has to be extended by the components of
the feed axis as described in [15], [17].

Figure 2: Mechanical model of the feed axis including ASU

Calculations in the time domain require the state-space description. The differential
equations of the system can be transformed to a state-space model as described in
[12]. This enables to implement the feed axis’ mechanical behaviour including the
® ®
ASU in Matlab /Simulink . To achieve a complete model of the test bench it is
necessary to extend the model by the control of the electric drive. Commonly, used
drives comprise of three control loops as described in [8], [23]. The inner loop,
current or force control consists of a conventionally tuned PI controller with only
minor couplings to the mechanical structure. The speed and position control rather
enclose a stronger coupling to the structure. Especially the point at which the
sensor is located plays a decisive role for the dynamics of the drive [3]. Hence the
model was enlarged by the three control loops including direct and indirect position
measurement. To simulate the system in the time domain it is furthermore
necessary to implement the limitations of the real drive as described in [1]. Besides
electrical parameters such as voltage and current, also the movement parameters
like jerk, acceleration, or speed have to be limited.

1.3 Control concept

The behaviour of a dynamic system is characterized by the position of the poles in
the complex pole zero pattern. Insertion of forces that are proportional to movement
parameters as for instance acceleration, speed, or position gives the chance to
change these poles. To clarify this context Figure 3(a) shows a one mass oscillator.
This system can be described by the following linear differential equation.
m&x&(t ) + dx& (t ) + cx(t ) = Fi (t ) (1)

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.

1.4 The feed axis test bench

To investigate a real feed axis system, a modular test bench was constructed,
assembled, and set into operation. Because of its modularity, the test bench offers
the possibility to change the ball screw drives and to vary the load (Figure 4).
Furthermore, it is equipped with a direct and an indirect position encoder to realize
the two common approaches of position detection in feed axis systems.

Figure 4: conceptual design of the modular feed axis test bench

The synchronous drive is connected to the spindle by a flexible coupling to avoid

direct backlash of axial vibrations to the motor shaft. The drive electronics is a
conventional position controller system with the commonly used three controller
cascades [8]. The controller parameters and several reference value curves can be
set by software. This software furthermore enables to record drive signals of
interest as for instance torque, speed, and position.
The combination of variable load and variable spindle pitch enables to reproduce
the most common feed axis configurations for miscellaneous kinds of machines or
production processes as described in [18]. In case of high cutting forces, those
appear for example in rough milling machines, a high load spindle with small pitch
being used, whereas with handling axis small load spindles and high pitches are
applied.
The ASU power supply is realized by amplifiers that where specially developed for
ohmic capacitive loads that are driven with very high frequencies. These amplifiers
where controlled by a differential voltage input signal. The output is directly
connected to the parallel piezo actuator circuits in differential setup.
®
The ASU control is realized by a dSpace rapid prototyping system whose software
enables free programming of the control algorithm. The controller is designed and
®
converted into executable real time code using Matlab/Simulink .

1.5 Verification at the test bench

In order to verify the model, the feed axis was firstly investigated without the ASU.
Therefore, the different configurations where assembled and the control was
optimized and aligned with the same measurement routine. Due to the difference
between position control using direct or indirect position detection [3], this routine
had to be accomplished twice for each configuration. The comparison of the
simulation and the measured results shows that the frequency behaviour, especially
the values of the first eigenfrequency of the model, fit to the values of the real
system configuration. The same procedure was repeated for the feed axis with
mounted ASU. Figure 5 (b) shows the position step response of the feed axis with
direct position detection as an example.
Step Response
120

Kv without Kv with Enhance-

100
Handling ASU [1/s] ASU [1/s] ment [%]
Indirect position measurement 8000 20000 150 80

direct position measurement 8000 14000 75

Position [µm]

60
Standard Milling

indirect position measurement 7000 11500 50 40

direct position measurement 11000 17000 60

20

Heavy Milling K=40000, without ASU

K=40000, with ASU
0
indirect position measurement 15000 40000 160 K=15000, without ASU
Reference Value
direct position measurement 10000 25000 150 -20
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time [s]
(a) Enhancement of Dynamics for different configurations (b) Step response of the feed drive
Figure 5: Mesurement results of the ASU operated in a feed drive

Regarding the achievable amplification in the position control loop (Kv-factor) it is

obvious that the ASU is able to improve the dynamic behaviour of the system. The
unacceptable tracking behaviour of feed axis with an excessive Kv (dotted line) can
be improved by using the ASU to rectify the overshoot. However, this potential is
limited by the maximum stroke of the ASU.
The measurement procedure was repeated for the other feed axis configurations. In
summary, the measurements show a significant increase of dynamics compared to
conventional feed axis when ASU is used.
1.6 Results
During the optimisation of feed axis considering control aspects the common
objective aims to achieve high dynamics and small contouring errors. An increase
of the position controller amplification Kv effects both destinations positively, but
rather reduces damping of the closed loop control [3]. Therefore, the amplification
Kv is limited. Consequently, it can be considered as an evaluation criterion for the
dynamics of the feed axis Figure 5(a) shows the achieved enhancements of Kv for
the investigated feed axis configurations. From the values it becomes apparent that
the best effects of using the ASU can be achieved either in the handling axis with
indirect position detection or in the axis of milling machines used for heavy
roughing.
Besides enhancement of the dynamics, also the costs for the overall system have
to be considered. That includes the encoders, the piezo based ASU, the control
hardware, and the piezo amplifiers. Especially the costs for amplifiers and actuators
can easily exceed the expected improvements. Heavy milling systems for example
require ASUs with huge applicable forces. The costs can therefore only be justified
by saving the costs for a direct position encoder. Handling systems on the other
hand are more suitable to be improved using ASUs. Feed axes for common milling
machines did not show similar improvements like heavy milling or handling axes.
The reason is related to the high resonance frequencies and stiffness that a priori
allow higher dynamics.

2 Adaptive spindle support for enhanced cutting

processes
A suitable approach to improve productivity is to operate multiple production steps
by one machine that is within the same clamping of the workpiece. Therefore, it is
necessary to develop machines that on the one hand offer a high chipping volume
for pre-machining production steps and on the other hand guarantee high precision
for the finishing production steps. Optimization of existing systems as described in
[14] is an appropriate way, but the estimated potential is rather limited. In contrast,
the redesign of whole systems using mechatronic components [16] provides
significant improvements. To overcome this conflict different approaches have been
established, with uni-axial approaches [21],[26] being the most relevant. However,
applying uni-axial approaches in multiple axes systems require increased efforts in
mechanical design and cause limitations concerning the stiffness. A multiple axes
interface was developed to overcome these problems. The mechanism consists of
six piezoceramic actuators in a parallel arrangement. While one end is fastened on
the main housing, the other end is connected to a solid state joint that supports a
standard HSC-spindle. As shown in Figure 6 there are three pairs of actuators
located around the z-axis designed in a differential configuration.
The mechanical design of this component was published in [5]. In comparison to
similar solutions [4], this mechanism not only compensates deformations of the tool
or damps vibrations, but is also suitable to realize defined deflections of the TCP.

Figure 6: CAD model of the adaptive spindle support (left) and real component (right) mounted in
a Parallel-Kinematic Machine (PKM).

2.1 Modelling and experimental analysis

The scheme of the adaptive spindle is shown in Figure 7. The vector Zr is derived
from the given surface geometry and is set as reference value of the control loop.
Due to the configuration of the piezo actuators the reference values have to be
transformed into the actuator coordinate system. The resulting vector contains the
necessary displacement of each actuator and represents the reference values for
six power amplifiers. The spindle position is measured by five capacitive
displacement sensors. Considering the geometric positioning of the sensors, the
displacement of the Tool-Centre-Point (TCP) in Cartesian coordinates has to be
recalculated via sensor transformation. To realize a control, the measured TCP-
displacements are compared with the reference values. The difference between the
reference values Zr and the actual values Za represent the controller input. The
controller calculates an actuating value for each of the Cartesian axes.
Figure 7: Control scheme of the adaptive spindle support.
The parallel kinematic design causes couplings between the different axes. A
reference displacement in x-direction for example, causes actuation values for all
actuators. Therefore, the dynamic behaviour of the system depends on the dynamic
behaviour in all directions. These couplings have to be considered in the designing
process of the controller.
To evaluate the mechanical design and to estimate the performance of the adaptive
spindle support, a FE-Model of the component was created. The support used for
the experimental investigation of the adaptive spindle was incorporated in the FE-
model to represent the boundary conditions realistically. Considering the piezo-
ceramic actuators as the dominant stiffness in the system the model not only
contains the mechanical properties of the component, but also represents the
electro-mechanical behaviour of the actuators. In order to take that into account the
simulation method presented in [12] was used. The method considers energy
conversion in piezoceramic actuators and expands the model by the electrical
parameters of the actuators. To compare the calculated model with the measured
data, a modal based state space model was created from the FE-Data. The inputs
of the state space model are the charges in the actuators and the forces in x,y,z-
direction at the TCP. The relative displacements of the structure at the sensor
points were defined as outputs. The model was exported in state-space-form and
extended by the necessary transformations. The FE-model was validated by
measuring the transfer functions from the actuators to the displacement sensors.
50
Gφx,φx measured
Magnitude [dB]

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.

2.2 Multiple-axes control concept

A linear MIMO-System (Multiple Input - Multiple Output) can be described in matrix
notation by the input vector U, the output vector Z and the matrix of the transfer
functions G. The notation is given in Eq. 8.

Z (s) = G (s) ⋅ U (s) (6)

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

The main objective in designing a decoupling network is to get a transfer behaviour

that only depends on the input signal of the main transfer path. This can be
achieved by adding an additional matrix between the controller output V and the
amplifier input U, as shown in Figure 10. The transfer function matrix of this
configuration is
Z (s) = G (s) ⋅ D(s) ⋅ V (s) (8)

To obtain a completely decoupled behaviour Eq. 8 has to be equal to

Z (s) = GD (s) ⋅ V (s) (9)

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.

The comparison of the coupled and decoupled transfer behaviour of a tilting

displacement around the ϕy-axis is shown in Figure 11 (a). Apparently, the
couplings resulting from the y-directed dynamics at 55Hz and 300Hz are almost
completely erased.

(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.

2.3 System validation during cutting operation

For tests the adaptive spindle support was mounted in the parallel kinematic test
bench 3POD as shown in Figure 12. The objective of the test was to produce a
finish in line boring in a 40mm pre drilled aluminium block with a feed amount of
th
0.1mm. The speciality of the borehole is a 4 order surface irregularity with an
amount of ±50µm. Therefore the adaptive component carrying the spindle is moved
in z-direction through the pre-drilled hole. Using a one-blade-cutter enables
production of a surface parallel to the xy-plane with a constant geometry in z-
direction. The reference values for the position control loop can be directly
calculated from the given surface. Due to the two-dimensional irregularity of the
surface, two sinusoidal functions for the x- and y-direction have to be determined.
The frequency of these functions depends on the rotary speed of the spindle.
The described geometry can be created by moving the spindle translational in x-
and y- direction or by tilting the spindle around x- and y- axes. Due to the mass of
the spindle an operation in the translational mode was only possible up to 40Hz. For
higher frequencies the performances of the actuators and amplifiers need to be
increased. The same operation in tilting mode allows much higher frequencies. To
examine the maximum operation frequency the measurements where repeated with
increased spindle speed.

Figure 12: Experimental setup (left) measured surface geometry (right).

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.

3 Compensation of thermal deformations in

machine structures using thermo sensitive
materials
In machine tools there are several heat sources that vary in time and position.
Combined with changes in the environmental conditions complex temperature
distributions occur. The resulting problems are varying thermal deformations
Common machines are optimized for using under defined conditions. If these
conditions differ, the thermal deformations cause a loss of accuracy as well as non
optimal drive conditions
A well distinguished and a common realized way to avoid such irregularities is using
the machines CNC-control unit to compensate thermal deformations. There are
different approaches for realization. For instance, it is possible to use a thermal
model of the machine to determine the thermal deformations as well as using
additional sensors. The basis of these approaches is to use the main drives to
compensate the deformations. This is an adequate and simple solution if the
deflection occurs in the actuating direction of the main drive. But it is rather complex
to use it to compensate multi-axial deformations or deformations that differ from the
main drives moving direction.

(a) Simulation results (b) mechatronic compensation device

Figure 13: Thermal deformations of a head stock and mechatronic compensation unit

A suitable way to deal with this problem is to design thermo-stable structures.

Figure 13 (a) shows exemplarily the thermal deformation of a head stock. The head
fixes the main spindle of the machine tool. Deformations on this part of the structure
directly effect to the Tool-Center-Point. The deviation of the TCP is multi-axial and
differs from the directions of the main drives. To compensate the deformation a
mechatronic device as shown in Figure 13 (b) was developed. The device consists
of a gear headed DC-drive and a mechanical flexure. The flexure represents the
interface between the drive and the structure. In combination with a temperature
sensor it is possible to react on different conditions. That offers the possibility to
compensate the deviation of the head stock and accordingly the deviation of the
TCP. Even though the device works seemingly good, the principle is not applicable
in all cases. There are limitations caused by the complex design, the limited forces
and the system costs. A recent approach to design thermo-stable or thermo-
adaptive structures is to use thermal Shape-Memory-Alloys (SMA).

3.1 Shape Memory Alloys (SMA)

3.1.1 Material Basics

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].

(a) mechanical principle (b) stress-strain behaviour

Figure 14: Common SMA-actuator design
The currently focused actuation approach is to use SMA-wires to design actuators
which replace electric drives as shown in Figure 15 [20]. Such wire actuators use
the contraction of the wire resulting by the phase transformation, to create
movements. Due to the required electrical energy and the system dynamics, this
approach is limited to actuators with small cross sections. Therefore the achievable
forces are less than 100N. The benefits of this approach are significantly smaller,
lighter and cheaper drives. Plenty of effort is spent in research to obtain a level that
allows realizing serial applications. Especially the automotive industry and the
consumer product industry push this work.

3.1.2 Structure-integrated SMA-Actuators

The demands to design thermo-stable structures using SMA’s cause completely

different requirements. Depending on the structure forces of several kilo-Newton’s
up to several ten kilo-Newton’s are necessary. The generated movements only
have to cover the range from several microns up to 1 mm. Therefore the actuator
geometry has to be changed completely from the wire to a more compact body.
Figure 15(b) shows an SMA-Actuator module which generates a movement of
0.9mm and a usable force of almost 7kN. This is realised by a cylindrical SMA-
Actuator of 10mm diameter and a free length of 40mm. The actuator is integrated in
an adapter which represents the interface between the actuator and the machine
structure. The force-displacement characteristics of the module can be adjusted by
choosing the diameter and the free length of the actuator that it meets the demands
of the structure. The necessary stiffness is created by disc springs parallel located
to the actuator body.

(a) SMA-wire actuator for automotive application (b) SMA-actuator-module

Figure 15: Different types of SMA-Actuators
There are two different operation modes for these actuators. The first is to use
external energy to heat the actuator. The application of electric current is not
possible due to the increased diameter of the actuator. It is rather possible to use
area oriented heat mechanisms like heating foils or electrical resistors mounted at
the surface. An alternative, but mostly more complex, approach is the application of
induction heating systems. However, in all cases the actuator is fully controllable.
Figure 16 (a) shows as an example a suitable system design of an adaptive ball-
screw drive.

(a) Fully controllable approach (b) Self-adapting approach

Figure 16: Concept of an adaptive ball screw drive
The design process for external heated actuators is focused on the mechanical
design and on the design of the heating and the sensing component. The thermal
characteristic of the structure is rather not very important for this operation mode. It
only has to be ensured that the heat energy, delivered to the actuator, does not
influence the surrounding structure. This can be realized by implementing some
layers or elements with a low heat conductivity to avoid a heat flow in the
surrounding structure.

3.1.3 Self-adapting Systems using SMA actuators

Another way to operate structural integrated SMA-actuators is to use thermal

energy which is accumulated by machining processes. Typical sources are friction
in bearings or the heat produced by electrical drives. This way is especially potential
if this waste heat causes thermal deformations which influence the machine
behaviour.
A possible system design is shown in Figure 16 (b). The heat produced by friction is
guided through the structure to heat up the SMA-actuator. When the actuator
reaches its transformation temperature, it will contract which causes a mechanical
feedback to the friction mechanism. In this operation mode the SMA-element
realises actuation function as well as sensing functions. According to the thermal
and the mechanical transfer path a closed loop, similar to a control loop, is created.
The closed loop is an inherent part of the structure. There is no need for external
control devices. That means that the system can be thermo-stable if it is designed
properly. Every heat input causes an actuator displacement which acts back at the
thermal deformation caused by the heat input.
In comparison to the active approach described in chapter 3.1.2 the thermal transfer
behaviour becomes more important. Due to the fact that the system behaves like a
control loop, the stability issue has to be focused in the design process. The
behaviour and the stability of the system can only be adjusted by designing the
mechanical structure. This involves the thermal behaviour of the structure, like
convection, conduction and the heat capacity, as well as the mechanical behaviour.
Of course the thermo-mechanic parameters of the actuator, like transformation
temperatures, stiffness an so on, are considered significant. Over all, very complex
interactions have to be considered during the design process. To understand these
interactions is one of the objectives of the project “Self-sufficient Self-Adaptive
Systems based on Shape-Memory-Alloys” within the Cluster of Excellenz
“eniPROD”.

3.1.4 Approach of an adaptive ball-screw-drive

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

3.2.1 Material basics

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

3.2.2 Approach of a thermo stable 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.

View publication stats