Escolar Documentos
Profissional Documentos
Cultura Documentos
1.
Introduction
Flywheel energy storage can be used in personal cars and commercial vehicles with a hybrid
propulsion system for either recuperating brake energy or for load leveling of the internal combustion
engine. These measures can reduce fuel consumption by up to 20 percent in urban driving cycles [1].
Flywheels offer high power density, a high number of cycles, no capacity fade over time and the state
of charge can be quantified exactly. Furthermore, no limited or environmentally harmful resources
need to be used [1, 2].
ICSV22, Florence, Italy, 12-16 July 2015
However, the flywheel bearings are still the limiting factor of this technology. Especially the high
requirements in terms of life span can hardly be reached by rolling element bearings. According to
the state of the art these requirements are only met by magnetic bearings, but their costs, their weight
and space requirement are significantly higher than in the case of rolling element bearings [3].
In order to reach the life time requirements the loads on the mechanical bearings in a flywheel
system would have to be reduced significantly. This work investigates the possibility to reduce the
effect of unbalance forces by applying an active vibration control strategy. Piezo-stack actuators are
used to counteract the unbalance forces. To carry out experiments, a test rig is built up and taken into
operation
2.
2.1
Fundamentals
Active vibration control systems are used to improve the dynamic characteristics of mechanical
structures, which are perturbed by vibrations. This is typically done by a closed loop control system. This system consists of sensors, measuring vibration induced quantities, a controller generating
appropriate corrective signals and actuators, controlled by these signals in a way that the unwanted
vibrations are reduced. Power amplifiers provide the power needed to drive the actuators.
Figure 1 shows a typical setup. The external disturbance force F excites a mechanical structure
with mass m and stiffness kS . The amplitude q of the induced vibration is measured by a suitable
sensor and fed to a controller. The controller operates the actuator in such a way that the force FA
reduces the disturbance force F .
Figure 1: Principle and components of an active vibration controlled mechanical structure with mass
m, taken from [4].
2.2
Commonly used actuators for active vibration control applications are electromagnetic, hydraulic
or piezoelectric-stack actuators [5]. The latter offer the possibility to generate high forces (up to 100
kN) and travel ranges applicable for vibration control. In addition, this type of actuator remains compact in terms of dimension, is suitable for highly dynamic applications and the achievable resolution
is in the range of a few nanometers [6].
Piezo-stack actuators are based on the inverse piezoelectric effect. This means, applying an electric field to a piezoelectric structure effects mechanical deformation of this structure. To achieve a
sufficiently large deflection, a multitude of thin piezo ceramic layers is stacked. Each layer is very
thin, so a quite low voltage suffices to achieve the required electric field strength [7].
2
If used for active vibration control, piezo-stacks can be treated like a spring with base excitation.
The spring constant is given by the stiffness of the actuator, the base excitation q0 corresponds to the
actuators travel range [4].
Figure 2: Active vibration controlled mechanical structure with a piezo-stack modelled as spring with
base excitation, taken from [4].
Figure 2 depicts the integration of a piezo-stack modeled as a spring with base excitation. The
resulting stiffness k 0 of the overall system is determined by the stiffness of the mechanical structure
and the stiffness of the piezo-stack. The equilibrium of forces can be written as follows [4]:
(1)
m q + k 0 q = F + k 0 q0
By controlling the base excitation q0 of the spring, which corresponds to the travel range of the
actuator, it is possible to modify the systems dynamic behavior and to counteract the vibration exciting force F . Current literature [4, 5, 8, 9] gives various different approaches for control strategies,
suitable for active vibration control applications with piezo-stack actuators. Generally, one can differentiate between model based (e.g. LQR control) and non-model based control strategies like velocity
feedback control [5].
2.3
Active vibration control can be applied to arbitrary mechanical structures. This work focuses
on the vibration control of a simple rotor-bearing system, consisting of a simplified flywheel-like
topology. The actuator is located at one of the bearing housings. A simple model for this setup is
shown in Fig. 3.
According to the contributions carried out in Section 2.2, the model consists of a spring with base
excitation, simulating the actuator. The shaft is modeled as a massless cantilever beam with length
l and stiffness kW . The rotor mass is concentrated in a point at position l1 . At this point, a force F
causes a vibration with the amplitude q. With the definition of the length ratio = ll2 , the system
stiffness kRES can be calculated [9]:
kRES =
(2)
1
2
+
kW kA
1
The equilibrium of forces for the system depicted in Fig. 3 is then defined by [9]:
(3)
m q + kRES q = F + kRES q0
Figure 3: Model for active vibration control of a rotor-bearing system with piezo-stack actuators,
taken from [9].
Equation (3) shows the effect that, the nearer the actuator is placed to the exciting force F , the
more effective the active vibration control works. To achieve a complete compensation of F (this
means q = 0 and q = 0) the actuator needs to provide the following travel range [9]:
xREQ =
(4)
2F
kRES
The required travel range xREQ in Eq. (4) is twice the size of the necessary base excitation q0 ,
because the actuator needs to counteract vibrations in both (negative and positive) directions.
In this section an overview of basic ideas of active vibration control is given. The following
sections deal with the experimental part of the work, namely with the designed system for measuring
the actuator deflection and finally the design and implementation of the test rig.
3.
A piezoelectric accelerometer is used to measure the acceleration of the free end of the piezostack. In order to determine the applicability of the chosen sensor actuator configuration for active
vibration control, the actuators travel range is measured. For this purpose, a measurement system is
designed and realized.
Accelerometers are the standard sensors for condition monitoring and vibration analysis of rotating machinery [10, 11] and therefore already integrated in existing flywheel systems. These sensors
are easy to handle, can operate in a wide frequency range and their sensitivity is independent from the
length of the measuring lead [12].
3.1
Measurement system
The accelerometer is directly mounted on the moving end of the piezo-stack. The actuator deflection is obtained by filtering and double integration of the sensor signal. To realize the measurement
system, a dSpace signal processor system in combination with Matlab/Simulink and dSpace ControlDesk is set up. Figure 4 depicts a schematic diagram of the system.
The sample frequency of the system is set to 20 kHz. The resonant frequency of the used unloaded
actuator is 14 kHz. If the actuator is fixed at one end, this value is halved [6]. In order to achieve a
correct behaviour of the actuator, frequency components lying above the resonant frequency have to
be filtered out of the driving voltage signal. This is done by an active Bessel reconstruction filter of
order 4 with a cutoff frequency of 3 kHz.
4
Determining the distance x(t) out of the measured acceleration a(t), a double integration of the
measurement signal has to be done. This is realized in Matlab, using a cumulative trapezoidal numerical integration method. Before integration, the measurement signal has to be bandpass filtered around
the adjusted frequency of the driving signal, to eliminate the direct component and high frequency
disturbances.
3.2
Results
The used piezo-stack P-842.20 from the company Physik Instrumente (PI) has a travel range of 30
m and a nominal input voltage range of 0 to 100 V, which can be extended up to -20 to 120 V without
damaging the actuator. For the presented results, the actuator was driven in the range of 0 to 100 V.
Driving the actuator with sinusoidal signals requires a bias voltage of half of the peak amplitude. This
effects a zero point offset of the travel range and allows actuator deflection in negative and positive
direction.
The actuator is driven with three different frequencies (50 Hz, 100 Hz and 166 Hz), therefore
different bandpass filters have to be designed. This is done in Matlab with the filter design tool fdatool. All filters are IIR-Butterworth bandpass filters with minimum order, the pass band is symmetric
around the driving signal frequency.
Figure 5(a) shows the unfiltered measurement signal for three different driving voltage amplitudes,
as well as the prescribed driving voltage.
USensor in V
UPiezo = 20V
0.02
UPiezo = 30V
UPiezo = 40V
8
6
0.02
4
0.96
0.97
0.98
t in s
Zeitverlauf Aktuator-Steuersignal
0.99
UPiezo in V
80
x A in m
0.04
0.95
2
0
2
4
60
40
20
10
0.95
UPiezo = 20V
UPiezo = 30V
UPiezo = 40V
10
0.96
0.97
0.98
0.99
t in s
0.95
0.96
0.97
0.98
0.99
t in s
signal.
The obtained results of the measurement system show, that the used sensor actuator configuration
is suitable for active vibration control applications.
4.
4.1
The design of the test rig follows the considerations presented in Section 2.3. Figure 6(a) shows a
schematic diagram of the test rig.
To increase the efficiency of the active vibration control, the actuated bearing is located near the
rotor disk (see also description of Eq. (3)). Two sensor actuator pairs are mounted in perpendicular
planes to provide active vibration control. The active bearing is shown in Fig. 6(b). The bearing
consist of two preloaded angular contact ball bearings in a face-to-face
arrangement. This setup eliminates the radial bearing clearance and therefore a more effective actuation can be achieved.
The rotor disk has 8 drill holes uniformly distributed around the circumference to attach unbalance
masses. The electrical drive offers the possibility to vary the speed in a range of 5000 rpm up to
25000 rpm. All components are mounted on a massive steel plate which is supported by rubber
buffers to minimize environmental influences. The measurement signal processing and control system
implementation is realized in Matlab/Simulink and dSpace ControlDesk in combination with a dSpace
signal processor system.
4.2
For the test runs, two non-model based control strategies are applied, velocity feedback and direct
acceleration feedback. The results shown in Fig. 7 are obtained by applying velocity feedback. This
control strategy is based on the idea of using the actuator as a controlled velocity proportional damper.
The adjustable control parameter is the gain of the damping factor.
The experiments are carried out with a rotational speed of 5000 RPM. Figure 7(a) shows how the
vibration orbit of the bearing decreases when active vibration control is applied. This reduction is
caused by the piezo-stack induced decrease of the bearing loads. In Fig. 7(b), the controlled actuator
driving voltage is shown. In order to avoid critical conditions in test operation, the controller gain
values are chose to be in a low range. This effects lower amplitudes of the generated actuator driving
voltages but of course also lesser vibration damping.
6
(a) Uncontrolled (blue) and controlled (red) vibration orbit of the active bearing.
5.
This work deals with the design and implementation of a test rig for active vibration control
with piezo-stack actuators. It could be shown that accelerometers in combination with piezo-stacks
provide a useful sensor actuator configuration for active vibration control of rotating machinery. Test
runs show the correct functionality of the test rig. By applying a non-model based control strategy, it
was possible to reduce the vibration orbit of the actuated bearing significantly.
For further test series the test rigs setup was revised, which is shown in Fig. 8. The frame builds
shorter than before and it allows the test bed to be flipped, so that the shaft axis runs vertically. This
will eliminate the influence of gravity in radial direction of the active bearing and offers the possibility
to simulate the shaft position of a mobile flywheel application. Different positions of the unbalance
mass can be implemented easily, since it is fixed on the shaft through grub screws. Different measurement equipment may be applied without redesigning the frame. The new design shows improved
stiffness and versatility. As a result of the increased stiffness the influence on results caused by the
setup will be reduced. Further the new frame provides modular expandability. Changes like shaftlength, type of bearings, mass of unbalance and installation of different piezoelectric actuators may
be executed without changing the whole frame and setup of the test rig.
The carried out work is a significant contribution to achieve the project goals of the ongoing
research project LESS (Lifetime increase in flywheel energy storage system), which is supported by
the Mobilitt der Zukunft program of the Austrian Research Promotion Agency FFG (Grant No.
FFG 845335).
ICSV22, Florence, Italy, 12-16 July 2015
REFERENCES
1 Bader, M., Buchroithner, A., Andrasec, I. and Brandsttter A, Schwungradhybride als mgliche Alternative
fr den urbanen Individual- und Nahverkehr, ENInnov2012-Proceedings, Graz, Austria 15-17 February,
(2012).
2 Schweighofer, B. and Wegleiter H., Design and Modeling of Mobile Energy Storage Systems, Lecture
Notes, Graz University of Technology, (2014).
3 Recheis, M., Buchroithner, A., Andrasec, I., Gallien, T., Schweighofer, B., Bader, M. and Wegleiter, H., Improving kinetic energy storage for vehicles through the combination of rolling element and active magnetic
bearings, Journal of energy & power engineering, 8, 1323-1332, (2014).
4 Ehmann, C., Methoden und Komponenten fr die Realisierung aktiver Schwingungsdmpfung, PhD Thesis,
Technische Universitt Darmstadt, (2004).
5 Roshdi, A. A., Robuste Regelung zur aktiven Schwingungsdmpfung elastischer Rotoren mit PiezoStapelaktuatoren, PhD Thesis, Technische Universitt Darmstadt, (2005).
6 Physik Instrumente (PI) Gmbh & Co. KG (2009). Piezoelectrics in Positioning, Tutorial on Piezotechnology
in Nanopositioning Applications. [Online.] available: http://www.pi-usa.us/pdf/index.php.
7 Wang, Q., Piezoaktoren fr Anwendungen im Kraftfahrzeug, Messtechnik und Modellierung, PhD Thesis,
Ruhr-Universitt Bochum, (2007).
8 Kaltesch, C., Vergleich einer piezoelektrischen Lagerabsttzung mit Quetschldmpfern zur
Schwingungsreduktion eines elastischen Rotors, PhD Thesis, Technische Universitt Darmstadt, (2011).
9 Horst, H. G., Aktive Schwingungsminderung an elastischen Rotoren mittels piezokeramischer Aktoren,
PhD Thesis, Technische Universitt Darmstadt, (2005).
10 Randall, R. B., State of the Art in Monitoring Rotating Machinery - Part 1, The Noise and Vibration Control
Magazine, March 2004, 14-20, (2004).
11 SKF (2015). SKF Vibration Sensors Catalog. [Online.] available: http://www.skf.com/binary/
82-48549/CM-P1-11604-EN-SKF-Vibration-Sensors-Catalog.pdf.
12 Glles, M., Vibrationsanalyse, Diploma Thesis, Graz University of Technology, (2003).