Turkish J. Eng. Env. Sci.32 (2008) , 51 – 57.c© TÜBİTAK

Active Control of Residual Vibrations of a Cantilever Smart Beam

Zeki KIRAL, Levent MALGACA, Murat AKDAĞDokuz Eylül University, Faculty of Engineering, İzmir-TURKEY

e-mail: zeki.kiral@deu.edu.tr

Received 04.12.2007

Abstract

The aim of this study is to control the residual vibrations of a clamped-free beam subjected to a movingload. The dynamic response of the beam is calculated by using the finite element method in order todesign a suitable control technique and the numerical results are verified by vibration measurements. Allthe numerical calculations are performed by a commercial finite element package. Two laser displacementsensors are used to measure the dynamic response of the beam. The moving load is obtained by pressuredair directed to the beam via a nozzle, and the movement of the load is achieved by an industrial robotmanipulator having 6 degrees of freedom. In this study, the suppression of the residual vibrations that occurafter the moving load has left the beam is considered as the main subject. Piezoelectric actuators are usedfor active vibration control study and displacement feedback is employed. The numerical results agree wellwith the experimental results. The results show that the finite element method can be used effectively fordesigning a suitable vibration control strategy.

Key words: Active vibration control, Finite element method, Moving load, Piezoelectricity

Introduction

Engineering structures work frequently under dy-namic excitations. The type of the dynamic exci-tation may vary but the results of these excitationsare shown generally in the form of vibration. Vi-bration can be attributed as an unwanted output formany engineering structures due to waste of energy,precision losses, noise etc. and should be kept un-der control especially for lightweight structures. Re-searchers still pay great attention to vibration sup-pression due to its industrial importance and thereare numerous studies related to active vibration con-trol.

Lim et al. (1997, 1999) investigated vibrationcontrollability of beams with piezoelectric sensorsand actuators using finite element analysis in bothfrequency and time domain. They showed the sup-pression of vibration amplitudes with both constantdisplacement and velocity feedback. The sensor re-sponse is examined when a unit voltage is appliedto the actuator. Celentano and Setola (1999) de-

veloped a simplified model of a beam-like structurewith a bonded piezoelectric plate by integrating theusual electrical with the finite element method andmechanical models with a RLC circuit.

Manning et al. (2000) presented a smart struc-ture vibration control scheme using system identi-fication and pole placement. System identificationis carried out in 3 phases: data collection, modelcharacterization, and parameter estimation. Input-output data are collected by stimulating the piezo-electric actuators with a square wave signal and mon-itoring the strain gage response. Negative velocityfeedback is used as the controller to reduce vibrationamplitudes. Gaudenzi et al. (2000) investigated thisproblem both experimentally and numerically withposition and velocity control approaches. The nu-merical simulation is developed with the finite ele-ment method based on an Euler-Bernoulli model.

Bruant et al. (2001) described the modeling ofbeam structures that contained piezoelectric deviceswith a simple finite composite element. A simplecantilever beam structure is studied. Six mechanical

51

KIRAL, MALGACA, AKDAĞ

degrees of freedom and 4 electric degrees of freedomare used in the model. They developed a method-ology for the determination of the optimal geome-tries of piezoelectric devices. Halim and Moheimani(2002) aimed to develop a feedback controller thatsuppresses vibration of flexible structures. The con-troller (H∞) is applied to a simple-supported PZTlaminate beam and it is validated experimentally.Yaman et al. (2002) described the role of smartstructures in aerospace engineering applications.

Kusculuoglu et al. (2004) developed a new FEmodel for a beam with a piezoceramic patch actu-ator. Each layer is treated as a Timoshenko beam.Two experimental studies are performed to validatethe theoretical developments. They observed thatthe use of the introduced model became more impor-tant when the piezoceramic and base layer thicknesswere large and shear and related rotational inertiabecame more important.

Karagülle et al. (2004) simulated the active vi-bration control of a cantilever beam having piezo-electric patches by ANSYS finite element package.They reported that modeling the smart structures,locating the actuators and sensors at the most suit-able positions, and determining the feedback gainis possible by commercial computer programs suchas ANSYS even for complicated structures. Xu etal. (2004) studied the effect of the sensor/actuatorlocation on the vibration control performance for aclamped-free beam using ANSYS. Malgaca (2007)studied the active control of free and forced vibra-tions of beam structures by piezoelectric actuators

both numerically and experimentally. Kıral et al.(2007) studied the moving load problem both experi-mentally and numerically and presented the dynamicresponse of a cantilever beam for different magni-tudes and velocities of the moving load.

In the present study, active control of residual vi-brations of a clamped-free beam subjected to a mov-ing load is considered. The moving load is a typeof dynamic excitation in which the position of theload varies in time. The vibration control problemis studied both experimentally and numerically. Theresidual vibrations of the beam that occur after theload has left the beam are suppressed effectively us-ing the piezoelectric actuators via displacement feed-back with zero reference input. The dynamic dis-placements during the moving load can be reducedby using suitable gain values for the proportionalcontroller. The selection of the controller gain can beachieved by programmable finite element programsin which the active elements can be employed. Themoving load mechanism used in this study has notbeen encountered in the literature and is the originalcontribution of this study.

Beam Model and Experimental Set-Up

In this study, a clamped-free beam having a rectan-gular cross section with dimensions 1.5 × 20 × 1000(mm) is considered. The material of the beam is alu-minum with the Young modules E = 69 GPa and thedensity ρ = 2676 kg/m3. The schematic view of theexperimental set-up is shown in Figure 1.

NI PCI-6722

V

air

Keyence LK-GLaser DisplacementSensors

KeyenceLK-G3001VLDM Controller

NI DAQPad-6015Multi FunctionalUSB DAQ Card

Analog OutputCard

SensorTech SS08ActuatorAmplifier

Piezoelectric actuatorsSensorTech BM532

A

B

Nozzle

100 mm

100 mm

xy

z

Pressured

Figure 1. Schematic view of the experimental set-up.

52

KIRAL, MALGACA, AKDAĞ

Piezo actuators

Figure 2. View of the experimental set-up, smart beam and piezoelectric actuators.

The force acting on the smart beam is obtainedby pressured air via a nozzle and the movement of theforce is established by a robot manipulator ABB IRB1400 shown in Figure 2. The robot is programmed tomove the nozzle with constant speed. The displace-ment response of the beam is measured via 2 laserdisplacement sensors located at the positions A andB shown in Figure 1. Two piezo patches located10 mm away from the clamped edge, SensortechBM532, are glued on the aluminumbeam and used asthe smart portion of the beam in monomorph config-uration as shown in Figure 2. The rectangular crosssection of a piezo patch has the dimensions 1 × 20× 25 (mm).

The measurement ranges of the laser sensors usedfor displacement measurement for point A and pointB are 150 ± 40 mm and 30 ± 5 mm, respectively.The sensitivities of the laser sensors A and B are 0.5µ and 0.05 µ, respectively. Dynamic displacementresponse recorded for point A is used as the feed-back for the vibration control study. The feedbacksignal is acquired via a LabVIEW (2003) programand then processed to form the control signal.

Experimental Results

Firstly, the natural frequencies of the smart beam aredetermined by free vibration analysis. Then 30 mminitial displacement is given to the free end of thebeam and the free vibrations of the smart beam aremeasured in terms of displacement via laser displace-ment sensors. The displacement signal measured bythe laser sensor B contains 3 fundamental frequenciesof the beam as shown in Figure 3. The fundamentalfrequencies are determined by fast Fourier analysis.

The load velocity is chosen as 0.25 m/s and theair pressure is regulated as 2.25 atm. Load velocityof 0.25 m/s is not a critical velocity and is chosenin order to observe the moving load response of thebeam for a sufficiently long time period. The mov-

ing load velocity can be changed in a broad velocityrange from 0.005 m/s to 4 m/s in the experimentalset-up. Figure 4 shows the step response of the beammeasured at point A to a 2.25 atm step load applied0.98 m away from the clamped edge. The distancebetween the nozzle and the beam surface is 25 mm,similar to the moving load case. The mean value ofthe steady-state displacement response is calculatedas –35.89 mm as shown in the figure and the loadvalue is calculated as 0.0427 N by the static deflec-tion expression

F =6E Iy uzs

x2 (3 a − x) (1)

where Iy denotes the inertia moments of area aboutthe y axis, uzs denotes the steady-state or static dis-placement at point A, x denotes the coordinate ofpoint A for which the static displacement is calcu-lated, and a denotes the position of the load. Theoscillations around the steady-state response can beattributed to the small changes in the air pressureand the solid/fluid interactions on the beam surface.

0 5 10 15 20 25-1

-0.5

0

0.5

1

Time (s)

0 5 10 15 20 25 300

0.5

1

Frequency (Hz)

Dis

plac

emen

t (m

m)

Mag

nitu

de

1.25 Hz

7.6 Hz

21.5 Hz

Point B

Figure 3. Free vibration response of the smart beam forpoint B and its frequency content.

53

KIRAL, MALGACA, AKDAĞ

Dis

plac

emen

t (m

m)

0 2 4 6 8 10 12 14 16 18 20-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

0

Time (s)

- 35.89 mm

Figure 4. Step response of beam measured at point A for2.25 atm.

The air nozzle is moved from the clamped end tothe free end of the smart beam by robot manipulator

and the displacement responses during the movingload acting on the beam and after the moving loadhas left the beam are recorded for control off andcontrol on cases. The closed loop block diagram ofthe active vibration control employed in this studyis shown in Figure 5.

As shown in Figure 5, the dynamic displacementresponse of the beam uz is used as a feedback sig-nal and compared with the reference signal, whichis 0 V, representing the case in which the beam isat rest. The error signal is processed by a propor-tional controller and the actuation signal is sent tothe piezoelectric actuators after amplification. Theamplification constant of the piezo amplifier is 30.The piezoelectric actuators drive the beam in orderto reduce the dynamic displacements. The upperlimit of the voltage value applied to the piezo actu-ators is ±270 V. Figure 6 shows the experimentaldynamic responses of the smart beam for control offand control on cases at 0.25 m/s moving load veloc-ity.

ControllerAmplifier forPiezoelectric

Actuator

PiezoelectricPatch

(Actuator)Beam

LaserSensor

Controller forLaser Sensor

-

+Ref = 0 V

Moving Load

uz

Smart Beam

Figure 5. Block diagram of the closed loop control system.

0 5 10 15-40

-30

-20

-10

0

10

20

30

Time (s)

Dis

plac

emen

t (m

m)

Kp = 1.5 Kp = 1.5

Experimental

Control OffControl On

Figure 6. Dynamic responses of the smart beam for point A, 2.25 atm, 0.25 m/s.

54

KIRAL, MALGACA, AKDAĞ

The moving load completes its travel on the beamin 4 s. The controller gain Kp is chosen as 1.5 forboth cases: moving load acting on the beam and theresidual vibrations. It is shown from Figure 6 thatthe closed loop control with proportional controllerand piezoactuators does not have a considerable ef-fect on the moving load response but vibration con-trol is very effective in suppressing the residual vi-brations for the selected controller gain value.

Simulation Results with ANSYS

The finite element method has been used for manyyears to examine the dynamics of structures. In thisstudy, a commercial finite element package ANSYS10.0 (2006) is used to simulate the vibration controlof the smart beam. The active portion (piezoelectricpatches) of the beam is modeled by Solid5 elementsand the passive portion (aluminum beam) is modeledby Solid45 elements as shown in Figure 7.

Z

1

X

Y

FEB 14 2008

11:27:55

ELEMENT

U

CPClamped Edge

Solid 5(Active portion)

Solid 45(Passive portion)

Figure 7. The finite element model of the smart beam.

An APDL (ANSYS Parametric Design Lan-guage) code is developed to create the finite elementmodel of the smart beam. Modal and transient anal-yses are performed via this code. The APDL codeenables us to perform a closed loop control for vi-bration suppression. The finite element model of thesmart cantilever beam consists of 456 elements forthe passive portion and 40 elements for the activeportion. Having obtained the finite element model,the natural frequencies of the smart beam are calcu-lated as given in the Table.

As shown in the Table, the experimental andnumerical values of natural frequencies are in good

agreement. The third step in the simulation is thedynamic response analysis with closed loop control.All these steps are carried out by the APDL code,a small part of which is given in the appendix.Rayleigh damping, [C] = α [M ] + β [K], with co-efficients α = 0.00462 and β = 0.00231, is used inthe dynamic analyses. The values of the dampingcoefficients are determined by a trial and error pro-cedure. These damping coefficients give the sameexperimental and numerical decrement values calcu-lated from the successive peaks in the free vibrationresponse. The movement of the load in the simula-tion is established in the APDL code by altering thenode at which the force is applied for each time step.For every substep in the transient analysis the vari-ation of the load is assumed to be linear. Figure 8shows the simulated dynamic responses of the beamfor control off and control on cases.

Table. First 3 natural frequencies of the smart beam.

Natural Frequencies (Hz)Experimental ANSYS

1st mode 1.25 1.26

2nd mode 7.60 7.82

3rd mode 21.50 21.68

0 5 10 15-40

-30

-20

-10

0

10

20

30

Time (s)

Dis

plac

emen

t (m

m)

Kp = 5 Kp = 1.5

ANSYS

Control OffControl On

Figure 8. Simulated dynamic responses of the smartbeam for point A, 2.25 atm, 0.25 m/s.

In the simulation, the controller gain Kp is cho-sen as 5 during the moving load acting on the beamand Kp is chosen as 1.5 after the load has left the

55

KIRAL, MALGACA, AKDAĞ

beam. Different gain values could not be used in theexperiments due to instrumental restrictions. Thesimulated actuation voltages applied to the piezo-electric actuators are shown in Figure 9. The sim-ulation results given in Figure 8 show that the dy-namic displacements during the moving load can bereduced slightly, but the residual vibrations are sup-pressed efficiently with active control. The finite el-ement analysis package ANSYS is used successfullyto design a suitable vibration control strategy. Thesimulation results agree well with the experimentalresults.

0 5 10 15 20 25-300

-200

-100

0

100

200

300

Time (s)

During moving loadDuring free vibration

Acu

tatio

n V

olta

ge (

V)

Figure 9. Simulated actuation voltage.

Conclusions

Active vibration control of a cantilever smart beamis considered both experimentally and numericallyin this study. A contactless moving load mechanismand contactless vibration measurement system are

used in the experimental study. The moving loadmechanism used in this study has not been encoun-tered in the literature. The simulation of the closedloop vibration control with displacement feedback isachieved by using a commercial finite element pack-age. The piezoelectric elements that actuate thebeam to suppress the dynamic response are usedin the experimental system and are modeled by acommercial finite element package for the numericalanalyses. Residual vibrations of the smart beam aresuppressed successfully by proportional control. Thesimulated and experimental results are in very goodagreement. The design of active vibration control ofmore complex structures can be achieved with theprogrammable finite element packages, which enableus to use active elements.

Nomenclature

a position of the load in x directionF load magnitudeIy inertia moments of area about y axisuz dynamic displacement in z directionuzs static displacement in z direction[C] damping matrix[K] stiffness matrix[M] mass matrixα mass proportional damping coefficientβ stiffness proportional damping coefficient

Acknowledgment

The authors express their special thanks to theScientific and Technological Research Council ofTurkey (TÜBİTAK) for giving support to the re-search project with project number 104M379.

References

ANSYS software ANSYS Inc., Canonsburg, PA,USA, (2006). (www.ansys.com)

Bruant, I., Coffignal, G., Lene, F. and Verge, M.,“Active Control of Beam Structures with Piezoelec-tric Actuators and Sensors: Modeling and Simula-tion”, Smart Materials and Structures,10, 404-408,2001.

Gaudenzi, P., Carbonaro, R. and Benzi, E., “Con-trol of Beam Vibrations by Means of Piezoelec-tric Devices: Theory and Experiments”, CompositeStructures,50, 373-379, 2000.

Halim, D. and Moheimani, R.O., “Experimental Im-plementation of Spatial H∞ Control on a Piezoelec-tric Laminate Beam”, IEEE/ASME Transactions onMechatronics,7, 346-356, 2002.

Karagülle, H., Malgaca, L. and Öktem, H.F., “Anal-ysis of Active Vibration Control in Smart Structuresby ANSYS”, Smart Materials and Structures, 13,661-667, 2004.

56

KIRAL, MALGACA, AKDAĞ

Kıral, Z., Malgaca, M. and Akdağ, M., “Experimen-tal Investigation of Moving Load Problem”, Pro-ceedings of 13th National Conf. on Machine Theory,Sivas, 2, 351-360, 2007. (In Turkish).

Kusculuoglu, Z.K., Fallahi, B. and Roston, T.J.,“Finite Element Model of a Beam with a Piezoce-ramic Actuator”, Journal of Sound and Vibration,276, 27-44, 2004.

LabVIEW (2003). LabVIEW user manual. NationalInstruments Corporation, (www.ni.com)

Lim, Y-H., Varadan, V.V. and Varadan, V.K.,“Closed Loop Finite Element Modeling of ActiveStructural Damping in the Frequency Domain”,Smart Materials and Structures,6, 161-8, 1997.

Lim, Y-H., Varadan, V.V. and Varadan, V.K.,“Closed Loop Finite Element Modelling of ActiveStructural Damping in the Time Domain”, SmartMaterials and Structures,8, 390-400, 1999.

Malgaca, L., “Integration of Active Vibration Con-trol Methods with Finite Element Models of SmartStructures”, PhD Thesis, Dokuz Eylül University,Institute of Natural and Applied Sciences, 2007.

Manning, W.J., Plummer, A.R. and Levesley, M.C.,“Vibration Control of a Flexible Beam with Inte-grated Actuators and Sensors”, Smart Materials andStructures,9, 932-939, 2000.

Xu, S.X. and Koko, T.S., “Finite Element Analy-sis and Design of Actively Controlled PiezoelectricSmart Structures”, Finite Elements in Analysis andDesign, 40, 241-262, 2004.

Yaman, Y., Çalışkan, T., Nalbantoğlu, V., Ülker,F.D. and Prasad, E., “Application of the SmartStructures in Aeronautics and Astronautics Engi-neering”, Proceeding of Conf. on Defence Technolo-gies, SAVTEK, 333-339, 2002. (In Turkish).

Appendix

A1. A sample part of the APDL code for active vibration control of smart beam.

/prep7/title, Cantilever smart beam model with discrete multiple!actuators for moving load analysis!Parameters for Smart Structurenpzt=1 ! Number of PZT : (0.No PZT, 1.A)l1=1000e-3 ! Length of Metal Beamb1=20e-3 ! Widthh1=1.5e-3 ! Heightl2p=25e-3 ! Length of PZTb2p=b1 ! Width of PZTh2=1e-3 ! Height of PZTl1s=10e-3 ! Length of space to locate PZT#1l2s=5e-3 ! Length of space to locate PZT#2nebx=100 ! Number of elements in X direction for Metalnepx=5 ! Number of elements in X direction for PZTnepy=4 ! Number of elements in y direction for PZTnebsx=2 ! Number of elements in X direction fornebsy=0 ! Number of elements in y direction for Space!-------------------------------------------------!Material Properties!-------------------------------------------------et,1,solid45et,2,solid5,3r,1,0,0,mz

/solu ! Modal Analysisd,nv0,volt,0d,nv,volt,0alphad,alpharbetad,betarantype,modal,newmodopt,lanb,6 solve*get,f1,mode,1,freq*get,f3,mode,3,freqFinish!-------------------------------------------------------------------/solu ! Transient analysis for moving load problemantype,trans,newoutres,all,allkbc,0!-------------------------------------------------------------------*if,contsel,eq,1,then*do,t,to+2*dt,ts,dt ! Active control after moving load*get,uza,node,ncen,u,z err=ref-ks1*uza va=kpf*kv*err *if,va,ge,vmax,then va=vmax

57