VIBRATION SUPPRESSION OF A CANTILEVER BEAM BY ACCELERATION FEEDBACK CONTROL AND ASSOCIATED STRESS REDISTRIBUTION

P.J. Roberts, M. P. Bayon de Noyer, S.V. Hanagud Georgia Institute of Technology, School of Aerospace Engineering, Atlanta, Georgia 30332-0150

ABSTRACT In most structures, fatigue critical areas are associated with regions of high stresses. Passive stiffening of structures usually displaces these high stress regions. Thus, for most applications, active vibration control is preferred. However, the question of what effect an active vibration control scheme involving a set of actuators will have on stresses arises. In this paper, we study the ability to use active structural damping by use of acceleration feedback control in ABAQUS and the effect of using a bonded actuator to reduce vibrations in a cantilever beam. We will also observe the stresses induced by an active vibration control system. Based on a low frequency approximation, we use ABAQUS to simulate the vibration reduction using the URDFIL and the DLOAD subroutines to read the acceleration at the free end of the beam, use acceleration feedback control to calculate the control force and then input an optimized control force. We observe both the closed loop vibration magnitude reduction and the redistribution of stresses in the area of the actuator.

INTRODUCTION In the last decade, piezoelectric actuators have received an increased amount of interest due to their high efficiency in the area of active vibration control. In structures, these actuators can induce forces and moments that are proportional to the voltage applied across them. A large number of designs of these actuators have been developed during the past few years. Recently, actuators using piezoceramic stacks have been developed for vibration control. Piezoceramic stacks have been used as force inducing actuators in truss elements (Preumont, 1992) and for vibration reduction in plates by placing the stack between a stiffener and the plate (Young, 1996). These actuators have also been implemented as moment inducing actuators by placing the stack within cutouts in stiff beams and plates (Redmond, 1997) or mounting the stack in an external assembly for active tail buffet alleviation (Hanagud, 1999).

2002 ABAQUS Users Conference

In most structures, fatigue cracks are associated with high stress regions and cyclic loading. When stiffeners are added to these areas, the stresses are not necessarily reduced but instead are usually displaced to adjacent areas. Hence the question of whether an active vibration control scheme involving a set of actuators will reduce stresses in the whole structure or create high stress areas in the vicinity of the actuators arises. Once the stress distribution around actuators is understood, an optimization of the actuators can be performed to abate such stress concentrations. Active and passive damping systems using piezoelectric elements have been modeled with finite element models (FEM) with good results (Varadan, 1996). FEM methods have been used to obtain stresses around the bonding edges of piezoelectric wafer elements bonded to a beam. This work has shown that stresses at the bond edges and in the structure in the vicinity of the piezoelectric elements can grow (Seeman, 1997). In this paper, we will study the ability to use structural feedback control in ABAQUS and the resulting local stresses in the structure resulting from a moment inducing externally mounted piezoceramic stack actuator. In this actuator, the control moments are achieved by placing the piezoceramic stack parallel to the controlled structure at a distance from its neutral axis and at a selected orientation. This design, called Offset Piezoceramic Stack Actuator (OPSA), has the following advantages over the more classical PZT wafers. First, the piezoceramic stacks use the direct d33 piezoelectric coefficient instead of the transverse d31 coefficient. Second, by stacking piezoceramic layers, forces and displacements are added together. Third, the offset distance enables the designer to induce larger control moments to the controlled structure. This third advantage of the offset piezoceramic stack actuator over the PZT wafer provides an advantage over the cutout concept as well since the lever arm can be larger than the thickness of the structure. Finally, with the added requirement that the active element, the piezoceramic stack, is easily removable while the mount is attached to the structure, the maintainability of the offset piezoceramic stack actuator is far superior to the PZT wafer, which, once bonded, cannot be easily removed from the controlled structure. The system studied in this paper is based on an experimental setup that was developed in the Structural Dynamics and Smart Structures Laboratory in the School of Aerospace Engineering at Georgia Tech. The specimen consists of a steel cantilever beam, which is 24-in long, 0.5 in. thick and 2-in wide, with an offset piezoceramic stack actuator (OPSA) bonded on the upper surface close to the clamped end. The OPSA is made of two steel blocks bonded to the cantilever beam with a piezoceramic stack, Physik Instrumente PI-830.10, clamped between the blocks. For control purposes, an accelerometer is located at the free tip of the beam. This setup is illustrated in

2002 ABAQUS Users Conference

Figure 1.

The OPSA induces a control moment in the beam while a transverse disturbance force is applied in the

center of the free end of the beam, using a shaker. During the experiment, the acceleration signal is fed to an analog to digital converter then processed through the controller. The control signal is passed through a digital to analog converter and then fed to the piezo amplifier. For stability and effectiveness, the controller is implemented using a fifth order Runge-Kutta numerical integration. In this paper, an ABAQUS simulation of active damping using an OPSA actuator with acceleration feedback control is investigated. The finite element model is used to determine the vibration magnitude reduction based on an optimal control force. We will also compare stresses in the beam without actuator with stresses in the closed loop system.

FINITE ELEMENT MODELING The finite element model was constructed in the ABAQUS version 5.8 simulation package. The beam portion of the model had the same dimensions as the experimental specimen. In the model, the beam portion is composed of 384, 20-node quadratic, second-order, isoparametric 3-D brick elements, which use reduced integration. This type of element was chosen because it gives the best results in bending. The elements which make up the beam were assembled in two layers with each element measuring 0.5 x 0.5 x 0.25, aligned in forty-eight rows with each row consisting of four elements. The beam is constrained at the base along the beam root edges. The material properties of the beam are given in Table 1 and are based on ASTM-A36 steel.

The model used Rayleigh damping based on a 1.5 % modal damping ratio for the first two bending modes. For Rayleigh damping, the damping matrix, [C], is a linear combination of the mass, [M], and stiffness, [K], matrices.
~ ~ [M] + [C] = [K ]

(1)

~ ~ and are evaluated using the following Equation for the first two bending modes: The values of

i =

~ ~ i + 2 i 2

(2)

In the above Equation, i and i are the natural frequency and damping ratio of the ith bending mode.

2002 ABAQUS Users Conference

For this study, two models were generated. The first model was the steel cantilever beam by itself as a reference. The second model was the closed loop model. For both cases, the first three bending modes in the 2-3 plane, as illustrated in Figure 2, and associated natural frequencies were computed for the system.

Beam FEM model without actuator The first model was of the cantilever beam without the actuator mounted. The first three natural frequencies are listed in Table 2 and the calculated Rayleigh damping are in Table 3.

For the stress study, the beam was driven by a harmonic tip load at the natural frequency of the first bending mode of the beam. This tip load was collocated with the sensor node. The amplitude of the transverse disturbance force was chosen to be 0.9 lbf (4 N) to insure that the small displacement assumptions of the analysis were satisfied while generating stresses that are larger than 10 percent of the ultimate strength. The resulting tip displacements are illustrated in Figure 3. The amplitude of the steady state displacement is 0.217 in (5.51 mm).

Closed Loop FEM models For the closed loop model, the OPSA was added to the FEM model. The end mount closest to the root of the beam is made of two layers of four brick elements similar to those used for the beam, each measuring 0.5 x 0.5 x 0. 5 while the other end mount contains two layers of two brick elements. In order to simulate the piezoelectric stack the model used springs and pressure forces. Four tetrahedral elements were placed on each opposing face of each mount. These elements each have one node outward from the base mount they are installed on. Then an 8-node brick element with a surface area for the distributed load of 5mm x 5mm is place at each free node of the four tetrahedral set per mount. This area corresponds to the cross-section area of the actual piezoelectric stack. The piezoceramic stack was simulated by four linear springs in parallel mounted between the opposing faces of the two mounts. The equivalent spring stiffness of the stack was calculated using Equation 3 (Bayon de Noyer, 2000), the stack properties in Table 4, and then divided by four.

ks =

AsYEs c

(3)

2002 ABAQUS Users Conference

The four spring elements were connected between the two opposing 8-node brick elements. The stack forces were simulated by applying distributed loads on the opposing faces of the 8-node brick elements. The first three natural frequencies are listed in Table 5 and the calculated Rayleigh damping are in Table 6.

For this model, active vibration control was achieved using acceleration feedback control (AFC). The Equations of motion of a system under AFC are as follows (Bayon de Noyer, 1999).

[M ]{x} + [C]{x} + [K ]{x} = {act }[G ][ c ]{} + {f } {} + [ c ]{} + [ c ]{} = { [acc ]{x} 1p }

(4a-b)

In the above Equations, {x} is the vector of degrees of freedom of the FEM model, {} is the vector of the p compensator coordinate, {act} is influence vector of the actuator,

acc is influence row vector of the

accelerometer, [c] is the compensator damping matrix and [c] is the compensator frequency matrix, [G] is the feedback gain matrix and {1p} is a vector of length p, with one for each entry, to account for the fact that all compensators are placed in parallel. For the cantilever beam with the model of the OPSA, the influence vector of the actuator is zero everywhere but at the eight nodes where the control forces are applied. At each of the nodes, the magnitude of the influence is As/8. For the nodes on the side closest to the clamped end, the influence coefficients are negative while for the other mount, the influence coefficients are positive. The influence vector of the accelerometer is zero everywhere but for the node at which the sensor is place where the influence is unity in the transverse direction 2. For this study, we will only consider the system when it is driven at the first bending frequency. In this case, we assume that the system has a single degree of freedom, which is the first bending mode. Hence, Equations 4.a-b can be reduced to the following system:
2 2 + 2 s s + s = a 1c + f 2 + 2 c c + c = a 2

(5.a-b)

In these Equations, and are the modal coordinates of the first bending mode of the structure and of the compensator; respectively; s, c, s and c are the natural frequencies and the damping ratios of the structure and the compensator, respectively; is the scalar gain applied to the feedback signal; and a1 and a2 are the modal influence parameters of the actuator and sensor, respectively.

2002 ABAQUS Users Conference

In order to compute the compensator parameters, the modal influence parameters need to be computed. Using the mass normalized mode shape from a frequency analysis performed in ABAQUS of the first mode, 1, the influence coefficients a1 and a2 are given by:
a 1 = {1 }T {act } a 2 = acc {1 }

(6.a) (6.b)

The influence coefficients can be computed from directly reading the modal displacements from the ABAQUS frequency analysis and then computed using Equations 6.a-b. They are given in Table 7.

The objective of the controller in this study is to minimize the displacement response of the beam due to the harmonic transverse load. Hence, we will use the optimal approach for the design of the acceleration feedback controller based on the minimization of the H2 norm of the closed loop receptance for a given control gain (Bayon de Noyer, 1998). For such a design, the gain of the controller is the design parameter and the frequency and damping ratio of the controller are given by:
c = s
c = 1 2 a 1a 2

(7.a) (7.b)

For this study, the feedback gain, , is chosen to be 2.5. This value is motivated by three conditions. First, for acceleration feedback control stability, a sufficient condition is that the product a1a2 is positive (Bayon de Noyer, 1999). Second, the larger the gain, the larger the amount of reduction that can be attained. Finally, we want the control scheme to be feasible. This latest condition is due to the fact that the amount of control force that can be generated by the stack is limited. This value for the gain satisfies the first condition, provides an order of displacement reduction of ten and was selected to generate the maximum control forces that can be actually obtained. The controller parameters are given in Table 8.

2002 ABAQUS Users Conference

In order to implement the controller within ABAQUS, the following procedure for the computation of the control force was employed using the subroutines URDFIL and DLOAD:

Read tip acceleration at the end of a time increment.

At the beginning of the next increment, the acceleration from the end of the previous increment is read into URDFIL.

Next Step

This acceleration is passed onto DLOAD which uses a sixth order Runge-Kutta numerical integration to solve Equation 5.b to find the current value of the controller state.

This controller state is used to compute the distributed force per the following Equation:
Fc = c 2

(8)

This process is iterated every time increment of 0.001 second. The time interval of the simulation is chosen to be five seconds in order to end the simulation well into the steady state regime. The results of this analysis are illustrated by Figure 4. Compared with the open loop, the displacements are significantly reduced and steady state is reached in about a second as seen in
Figure 5.

The amplitude of the steady state vibration is 1.89 10-2 in (0.48 mm)

which is more than an order of magnitude smaller than the amplitude of the beam alone.

2002 ABAQUS Users Conference

STRESS STUDY For this study, in the case of the beam alone, the stress fields are obtained at a maximum positive displacement of the free tip of the beam. This choice is motivated by the fact that stresses are the largest at maximum displacement. The longitudinal stresses, 33, are illustrated in
Figure 7. Figure 6

while the shear stresses, 23, are illustrated in

The maximum longitudinal stresses are located on the upper surface of the root of the beam and have a magnitude of about 12 ksi (82.74 MPa), which is more than 10 percent of the ultimate strength of the material of the beam. Similarly, the maximum shear stresses are located in the same area and have a magnitude of about 4.3 ksi (29.65 MPa).

For the closed loop stress study, two different times should be considered. As for the beam alone case, the stresses at maximum displacement should be studied. However, since the control force have a 90-degree phase difference with respect to displacement, the largest control forces are exerted when the displacement is zero. Figure 8 and
Figure 9

illustrate the longitudinal and shear stress fields, respectively, for the maximum displacement case. At

this time, the maximum longitudinal stresses are still located on the upper surface of the root of the beam with a magnitude of 1.25 ksi (8.6 MPa), which is an order of magnitude less than the beam alone. Further, the longitudinal stresses in the vicinity of the OPSA are much smaller than the stresses in the same area for the beam alone. Similarly, the maximum shear stress magnitude is 0.55 ksi (3.8 MPa) on the upper surface of the root of the beam. These closed loop stresses are about an order of magnitude lower than the ones of the beam alone.

During the steady state, the maximum distributed control forces are obtained when the beam displacement is zero and their magnitude is about 5.8 ksi (40 MPa), which for the area of the stack is equivalent to a control force of 1 kN. This result validates our choice for the controller gain since it was designed to generate the maximum control force that could be delivered by the stack, 1 kN. When, the control forces are maximum, the highest longitudinal stresses within the beam are located in between the two mounts, as illustrated by
Figure 10.

The amplitude of these

maximum stresses is about 1.5 ksi (10.3 MPa), which is still an order of magnitude smaller than the maximum stresses in the beam alone and over six times smaller than the open loop stresses at the same locations.

2002 ABAQUS Users Conference

CONCLUSION For this study, we have developed an ABAQUS model was used to simulate active damping in a cantilever beam. This simulation has shown that the stresses within the system are reduced significantly when the displacements are reduced by an OPSA actuator. This simulation indicates that a control system based on OPSA actuators and acceleration feedback control is a feasible solution for fatigue life enhancement. Further, with this study we see how ABAQUS can be used for active structural control by using a combination of the URDFIL and DLOAD subroutines. Further studies into this closed loop stresses could explore an actual piezoceramic element mounted in the model instead of using springs. The actual contact/non-contact behavior could be studied. Other types of control schemes should also be considered for active structural damping. REFERENCES (1) Preumont, A., Dufour, J.P., and Malekian, C., Active Damping by a Local Force Feedback with Piezoelectric Actuators, J. of Guidance, Control and Dynamics, Vol. 15, No. 2, 1992, pp. 390-395. (2) Young, J.W., and Hansen, C.H., Control of Flexural Vibration in Stiffened Structures using Multiple Piezoceramic Actuators, Applied Acoustic, Vol. 49, No. 1, 1996, pp. 17-48. (3) Redmond, J., and Barney, P., Vibration Control of Stiff Beams and Plates Using Structurally Integrated PZT Stack Actuators, J. of Intelligent Material Systems and Structures, Vol. 8, 1997, pp. 525-535. (4) Hanagud, S., Bayon de Noyer, M., Luo, H., Henderson D., and Nagaraja, K. S., "Tail Buffet Alleviation of High Performance Twin Tail Aircraft using Piezo-Stack Actuators", Collect Tech Pap, Struct. Struct. Dyn. Mater. Conf., 1999, AIAA-99-1320. (5) Varadan, V.V., Young-Hun, L., Varadan, V.K., Closed Loop Finite-Element Modeling of Active/Passive Damping in Structural Vibration Control, J. of Smart Materials and Structures, v. 5, n. 5, 1996, pp. 685-694. (6) Seeman, W., Wolf, K.D., Straub, A., Hagedorn, P., Chang, F., Bonding Stresses Between Piezoelectric Actuators and Elastic Beams, Proceedings of SPIE, v. 3041, 1997, pp. 665-675. (7) Courant, R. and Hilbert, D., Methods of Mathematical Physics, Vol. 1, Interscience, New York, 1961, p. 277. (8) Hibbitt, Karlsson & Sorensen, Inc. ABAQUS/Standard User Guide, Version 5.8, 1998. (9) Bayon de Noyer, M., and Hanagud, S, Single Actuator and Multi-Mode Acceleration Feedback Control, J. of Intelligent Material Systems and Structures, v 9, n 7, 1999, pp. 522-545. (10) Bayon de Noyer, M., and Hanagud, S., Comparison of H2 optimized design and cross-over point design for acceleration feedback control. Collect Tech Pap, Struct. Struct. Dyn. Mater. Conf., v 4, 1998, pp. 3250-3258, AIAA-98-2091. (11) Bayon de Noyer, M., Tail buffet alleviation of high performance twin tail aircraft using offset piezoceramic stack actuators and acceleration feedback control, Thesis (Ph. D.)--School of Aerospace Engineering, Georgia Institute of Technology, 2000.

2002 ABAQUS Users Conference

Table 1. Beam material properties 29.0x106 Youngs modulus Poissons ration 0.3 7.355x10-4 Density
lbf sec 2 in 4

lbf in 2

200 GPa 0.3

7860 kg/m3

Table 2. First three natural frequencies of the FEM beam alone (bending in 2-3 Plane) Mode 1 Mode 2 Mode 3 27.808 Hz 174.01 Hz 486.07 Hz

Table 3. Rayleigh damping values for beam alone ~ = 4.5194

~ = 2.3658 10-5

Table 4. Properties of the piezoceramic stack (PI-830.10) Length, c Cross Sectional Area, As Youngs Modulus, YEs Piezoelectric Constant, d33 Density, s Number of Layer, ns 18 mm 25 10-6 m2 55.25 GPa 635 10-12 m/V 8000 kg/m3 ~ 200

Table 5. First three natural frequencies of the FEM beam with mounts and springs bending in the 2-3 plane Mode 1 Mode 2 Mode 3 29.823 Hz 181.48 Hz 495.21 Hz

10

2002 ABAQUS Users Conference

Table 6. Rayleigh damping values for beam with mounts ~ = 4.8280

~ = 2.2596 10-5

Table 7. Modal influence coefficients of the OPSA and accelerometers a1 = - 2.175 10-3 a2 = 15.27

Table 8. Controller parameters c = 187.38 rad/s c = 14.408 % = - 2.5

2002 ABAQUS Users Conference

11

Figure 1. Offset Piezoceramic Stack Actuator mounted on a cantilever beam

Figure 2. Wire frame of the FEM of the beam with OPSA

Displacement of Beam Free Tip without OPSA Installed 0.25 0.2 0.15 Free Tip Displacement (inches) 0.1 0.05 0 -0.05 -0.1 -0.15 -0.2 -0.25 0 0.5 1 1.5 2 2.5 3 Time (seconds) 3.5 4 4.5 5

Figure 3. Displacement of the free tip of the beam alone

12

2002 ABAQUS Users Conference

Displacement of Beam Free Tip with OPSA Installed 0.025 0.02 0.015 Free Tip Displacement (inches) 0.01 0.005 0 -0.005 -0.01 -0.015 -0.02 -0.025 0 0.5 1 1.5 2 2.5 3 Time (seconds) 3.5 4 4.5 5

Figure 4. Closed loop displacement of the free tip of the beam with OPSA and acceleration feedback control

0.2

0.15

0.1

Displacement (inches)

0.05

-0.05

-0.1

-0.15

-0.2

3 Time (seconds)

Figure 5. Open/Closed loop displacement of the free tip of the beam with OPSA and acceleration feedback control

2002 ABAQUS Users Conference

13

Figure 6. Zoom of central cross section of the longitudinal stress, 33, for the beam alone

Figure 7. Zoom of central cross section of the shear stresses, 23, for the beam alone

Figure 8. Zoom of central cross section of the longitudinal stress, 33, for the closed loop system at maximum displacement

14

2002 ABAQUS Users Conference

Figure 9. Zoom of central cross section of the shear stresses, 23, for the closed loop system at maximum displacement

Figure 10. Zoom of central cross section of the longitudinal stress, 33, for the closed loop system at maximum control force

2002 ABAQUS Users Conference

15