Experimental flutter and buffeting suppression using piezoelectric actuators and sensors

Transcription

1 Experimental flutter and buffeting suppression using piezoelectric actuators and sensors Afzal Suleman*a, Mtónio P. Costab, Paulo A. MOfljZa ad1\4fciflstftto Superior Técmco, Dept. Mech. Eng., Lisbon, Portugal baeronautical Laboratory, Portuguese Air Force Academy, Sintra, Portugal ABSTRACT This experimental investigation focuses on the application of piezoelectric sensors/actuators for wing flutter and vertical tail buffet suppression. The test article consists of a foam airfoil shell enveloped around an aluminium plate support structure with bonded piezoelectric actuators and sensors. Wind-tunnel test results for the wing are presented for the open- and closedloop systems. Piezoelectric actuators were effective in suppressing flutter and the wake-induced buffet vibration over the range of parameters investigated. Keywords: Wing, Flutter, Buffeting, Piezoelectric, Actuators, Sensors, Wind-tunnel, Experiment. 1. INTRODUCTION The advancement of technology in the search for multifuinctional materials has resulted in the concept of adaptive laminated composites. These electro-magneto-thermo-mechanical materials have presented an exceptional promise when compared to conventional ones. For deformation of thin structural elements, the most widely used multifunctional materials are piezoelectric actuators. Piezoelectrics have higher bandwidths than are possible in shape memory alloys, they are more compact than magnetostrictive devices and they are bidirectional by nature unlike electrostrictive materials. The piezoelectric behaviour can be manifested in two distinct ways. The 'direct' piezoelectric effect occurs when a piezoelectric material becomes electrically charged when subjected to a mechanical stress. As a result, these devices can also be used to detect strain, movement, force, pressure, and vibration by developing appropriate electrical responses. The 'converse' piezoelectric effect occurs when the piezoelectric material becomes strained from being placed in an electric field. The ability to induce strain can be used to generate a movement, force, pressure or vibration through the application of a suitable electric field. The use ofadvanced composites in aircraft structures has led to the development ofthe adaptive wing concept. This concept utilizes the properties of multifunctional materials in controlling the structural stiffness and shape of the composite materials in order to tailor aeroelastic response. Several investigators have studied the aeroelastic behaviour of composite plate wings. Albano and Rodden' used a doublet-lattice method to calculate lift distributions on wings in subsonic flow. Nam et al.2 used self-sensing PZT actuators for flutter suppression of a plate model of a wing. They also predicted the electric power required for aeroelastic control. Lazarus, Crawley and Lin3 showed the ability of both articulated control surfaces and piezoelectric actuators to control the dynamic aeroelastic systems. Scott and Weisshaar" examined the capability of panel flutter suppression with piezoelectric actuators and shape memory alloy actuators. Leeks and Weisshaar5 examined the interaction between the piezoelectric actuators and the stiffness of the host structure to produce large bending deflection. Ehiers and Weisshaar6'7 showed the effects of the piezoelectric actuators on the static aeroelastic behavior such as lift effectiveness, divergence, roll effectiveness. Very few experimental studies with the objective of demonstrating the feasibility of applying piezoelectric actuators and sensors in wing flutter and vertical-tail buffeting suppression have been documented. These include the research reported by Heeg8'9 to study a two-degree-of-freedom wind tunnel model. The purpose of this analytical and experimental work was to conduct a simple investigation of flutter suppression using piezoelectric patches. The test article had two primary flexible modes: a plunge mode and a pitch mode. Closed loop flutter tests exhibited a 20% improvement in flutter characteristics relative to the open-loop case. Other experimental research in aeroelastic response tailoring has been reported by Lin and Crawley' with the objective of demonstrating the ability of strain-actuated adaptive wings to suppress flutter and perform * Correspondence: suienan(daltäist.ud.pt, Tel: ; Fax: Part of the SPIE Conference on Industrial and Commercial Applications of Smart Structures 72 Technologies Newport Beach, California March 1999 SPIE Vol X/991$1O.OO

2 trade-off studies between the aerodynamic and structural actuators. Yet another experimental wing flutter suppression project has been reported by Heeg" using neural networks. In the area of active buffet load alleviation, the first experiment demonstrating the feasibility of using piezoelectric actuators for this purpose was also conducted by Heeg et al'2. A rigid wing was placed upstream ofthe test section to generate a wake flow representative ofa wing-horizontal tail configuration. The objective of this paper is to present the experimental vibration, buffeting and flutter response measurements made on a wind-tunnel model that was equipped with an active control feedback system using piezoelectric actuators and sensors. Damping and dynamic response data are presented for two cases: open loop and closed ioop with a controller based on an optimised feedback gain. The fundamental goal is to demonstrate experimentally the feasibility of using adaptive materials technology to suppress wing flutter and to alleviate buffeting. 2. WIND TUNNEL MODEL A photograph and a sketch ofthe wind tunnel set up and test article are presented in Figure 1. The wing model consists of a NACAOO12 airfoil enveloped around a rectangular aluminium plate structure. The rectangular planform wing has a 240 mm span and 140 mm chord. The airfoil shape was made of foam material and this was glued to the aluminium support structure. These dimensions were determined based on the wind tunnel size, blowing air velocity and the limitations ofthe piezoelectnc actuators. Twelve piezoceramic actuator patches (38x 25x 0.2 mm ) are bonded to the top and bottom ofthe plate near the cantilevered end. The actuators are configured to impart moments to the aluminium plate. Two piezoelectric sensor patches were bonded to the plate to monitor loads on the model. The output signal from the sensors was made proportional to the vertical displacement of the model. The sensor signal was sent to digital signal processor (DSP) through filters. The control signal was sent to power amplifiers. Amplified signal drove the piezoceramic actuators and attenuated vibration ofthe wing. The two configrations for buffet alleviation and flutter suppression are presented in Figures 2 (a) and (b). For buffet alleviation, a rigid wing was placed upstream ofthe test section to generate a wake flow representative ofwing-horizontal tail configuration, as opposed to the flutter suppression setup, where a clean upstream flow was assumed. 3. CONTROLLER DESIGN The controller design was based on the finite element structural model of the wing and a modal analysis was performed to generate natural frequencies and mode shapes. The vibration mode shapes and frequencies are shown in figures 3(a-d). Numerical models using the finite element method were developed and analyzed to determine the optimal placement of the piezoelectric actuating plates for maximum control-structure interaction. A block diagram of the controller is presented in Figure 4. The output analog signal from the PZT sensor mounted on the support root was sent to an analog-to-digital converter with a sample rate of 1O seconds. The PZT sensor signal is proportional to and in phase with the displacement of the model. The digitized signal was then sent to the control law that was implemented on a MATLAB-SIMUL1NK environment running under a real time operating system. The analog-to-digital converts were 32 bit units. The controller design in based on a simple feedback gain law. The gained signal was next sent to a one step time delay 1O s. The time delay provides a means for changing the phase of the feedback signal. The gained- and phased-shifted signal was converted back to an analog signal by a zero-order-hold digital-to-analog converter. The converted signal was connected to two operational amplifiers. The output signal from the amplifiers were used to drive the piezoelectric actuators. The maximum output voltage of this amplifiers was 100 V. For the present study a sampling rate of 1000 samples per second was used. This relatively high rate was chosen to ensure that the buffeting waveform is well defined. 4. EXPERIMENTAL PROCEDURE AND RESULTS The experimental set-up is based on a DSP state of the art laboratory facility created for testing and validating the theoretical models and application of active control methodologies. The vibration control test was first conducted to validate the results 73

3 obtained and these were compared with numerical results.. The goal was to improve the damping characteristics of the structure. The gain control method ustd for the vibration control was calculated according to the following equation: UKg G1 Y Here U is the Laplace transformation of input u, Kg S the multiplicative gain value, Gfis a transfer function for the low-pass filter, Yis the Laplace transformation ofthe outputy from the sensor. The addition ofthe first-order low-pass filters was to improve the quality ofthe sensor signal. Figure 5(a) shows the natural damped vibration response excited with a sine wave input through the piezoelectric actuators (controller off). Figure 5(b) corresponds to the actuator input signal which was turned off at t = 30 s.. The response in Figure 5(c) corresponds to controller active case, the time for the wing to stop oscillating being reduced from 37 seconds to 1.5 seconds. This simple test gave a good indication of the amount of damping produced by the actuators and control system and the amount ofpower required to achieve acceptable damping. The schematics for the buffet control test are the same as for the vibration control test, the only difference, is that the wing panel was placed inside a wind tunnel. The system has a single output and a double input, one for each group ofactuators. A rigid obstacle was placed at the upstream end ofthe test section to simulate wake flow emanating from a wing-horizontal tail configuration. The position of this object could be easily adjusted so that the resulting wake would impinge on the model mounted downstream. The configuration finally selected was the one that produced the largest buffet response of the model.. Preliminary tests were carried out with only the aluminum support plate inside the wind tunnel. The speeds at which the buffet would occur were around 5.2 m/s. With the control law gain set to the desired value the tunnel speed was increased to and then held constant at a pre selected value. Buffet response measurements were made at a velocity of 5.2 rn/s. The output signal from the PZT sensor was routed to a transfer function analyzer that was used to calculate the auto-correlation function of the response signal. With the controller on, it was observed that the buffeting vibration decreased by 30 %. Figure 6 (a) shows the displacement ofthe wing tip with the controller on until t 40s and then turned off It can be clearly observed that the buffet vibration increases considerably with the controller turned off Figure 6 (b) shows the corresponding signal sent by the controller to the actuators to minimize the vibrations. The setup for the flutter control test is the same as for the buffet control test, the only difference, is that the wing panel isto be inside a clean wind tunnel without any upstream obstacle. The turbulence within the tunnel was sufficient to perturb the model. Preliminary tests were carried out in order to determine the speed and angle of attach at which the flutter would occur. The onset offlutter varied from 30m/s at zero angle ofattach to 10 m/s at io angle ofattack without flow separation. The open and closed loop flutter tests were conducted by increasing the speed until flutter occurred. In order to validate the results, two additional tests were performed. The first test demonstrated that once flutter was encountered, turning on the controller was not sufficient to damp the motion. The second test illustrated that when the control law was turned off above the open-loop flutter velocity, divergent oscillations immediately began. With the control law gains set to optimal values, flutter response measurements were made at different velocities and AOA. The use of the feedback system resulted in the flutter speed increments shown on Table 1 for two angle of attack settings, two different sensor locations and two different control methodologies. For the Single Input Single Output (SISO) methodology a pair of sensors (one each side ofthe wing) were used near the root and two groups of actuators acting at the same time. For the Double Input-Double Output methodology (DIDO) two groups of actuators and sensors were utilized, one for the leading edge and other for the trailing edge, each group using its own feedback control loop. SLNGLEINPUT/ DOUBLE INPUT/ CONTROL SINGLE OUTPUT DOUBLE OUTPUT 00 loo OFF Vf Vf ON G % 44% 5.7% 6.03% 3.3%... 74

4 The wing model rigidity augmentation that resulting from bonding the piezoceramic transducers to the wing plate resulted in a 10% flutter speed increase and the performance was measured relative to this setting. Based on the experimental tests, it was also observed that the piezoelectric actuators exhibited a diminishing control authority as the air speed was increased. The inhibited control authority was due to the continued increase in aerodynamic loads as the airspeed is increased. 5. CONCLUDING REMARKS Wind tunnel tests have been conducted to demonstrate the feasibility of applying piezoelectric actuators and sensors to vibration control, buffeting alleviation and flutter suppression problems. In vibration control tests, the time taken to control the model was 1.5 seconds for the controller design implemented in this study. In active buffeting attenuation, a 30% reduction in buffeting was observed. In the case of the flutter suppression, a 6% increase in the critical flutter speed was attained. Further analysis and experiments using this model are under investigation for closed loop wind tunnel testing. ACKNOWLEDGEMENTS The authors wish to acknowledge the financial support received from the Fundação para a Ciencia e Tecnologia in the framework ofproject PRAXIS/3/3. 1/CEG/265 1/95. REFERENCES 1. E. Albano., W.P. Rodden, "A Doublet-Lattice Method for Calculating Lift Distributions on Oscillating Surfaces in Subsonic Flows", AIAA Journal, Vol. 7, No. 2, Feb. 1969, pp C. Nam, S. Oh, W. Kim, "Active Flutter Suppression ofcomposite Plate with Piezoelectric Actuators", AIAA Paper No , Proceedings ofihe AJAA/ASMEAdaptive Structures Forum, SC, April KB. Lazarus, E.F. crawley, c.., Lin, "Fundamental Mechanisms of Aeroelastic control with control Surface and Strain Actuation", AIAA Paper No , Proceedings of the 32nd Structures, Structural Dynamics and Material Conference, MD, April R.C., Scott, TA. Weisshaar, "controlling Panel Flutter Using Adaptive Materials", AIAA Paper No , Proceedings ofthe 32nd Structures, Structural Dynamics andmaterial Conference, MD, April T.J. Leeks, TA. Weisshaar, "Optimizing Induced Strain Actuators for Maximum Panel Deflection", AIAA Paper No , Proceedings ofthe AJAA/ASMEAdaptive Structures Forum, S, April SM. Ehiers, TA. Weisshaar, "Static Aeroelastic Behavior of an Adaptive Laminated Piezoelectric composite Wing", AIAA Paper No , Proceedings ofthe 31st Structures, Structural Dynamics andmaterial Conference, ca, April SM. Ehlers, TA. Weisshaar, "Effects of Adaptive Material Properties on Static Aeroelastic control", AIAA Paper No , Proceedings ofthe 33rd Structures, StructuralDynamics andmaterial Conference, TX, April J. Heeg, A. McGowan, E. crawley, C. Lin., "The Piezoelectric Aeroelastic Response Tailoring Investigation: A Status Report" Proceedings ofthe SP1E Smart Structures and Materials, San Diego, ca, J. Heeg, "Analytical and Experimental Investigation offlutter Suppression by Piezoelectric Actuation", NASA TP-3241, Feb C. Y. Lin, E. F. crawley, "Strain Actuated Aeroelastic control", SERc#2-93, Massachusetts Institute of Technology, February J. Heeg, "An Analytical and Experimental Investigation offlutter Suppression Via Piezoelectnc Actuators", AIAA Paper No , Proceedings of the AIAA Dynamics Specialist conference, April J. Heeg, J.M. Miller, and R.V. Doggett, "Attenuation of Empennage Buffet Response Through Active control of Damping Using Piezoelectric Material, NASA JM , February

5 WINT TUNNEL Figure 1 - the experimental setup and schematic 76

6 Figure 2 (a) - Test article in the wind-tunnel with the upstream obstacle to generate wake flow for buffeting alleviation Figure 2 (b) - Test article in the wind-tunnel in a clean setup to suppress flutter 77

7 03 tt ci. Cd, 0 3 C '1 a Cd) 0 a hg. niinnnnawwsanawn.. : : nnnnan a fl r1 w flwflctflflflfl* T k W*fl%tO P flt4flflflmntw fl b$ fl fl M I S U n NflCflPflflIflfl 1a '... n*can*nnwna M**s*flflflgfl*flflflsfl

8 DS2603BI Demux Sine Wave Gaini Muxi DS2H)3R1 Constant Figure 4- Schematic of the controller design using MATLAB/SIMULINK 79

9 I [\ /\ j A (a) I I I \ / \ H 1k 1 \ I ' I \ I \ \j i IJ ' \\ Si Figure 5 - Experimental vibration control results: (a) - Natural damped vibration response with controller off; (b) Actuator input signal which was turned off at t= 30s; (c ) Controlled response showing attenuation within 1.5 s. 80

10 Figure 6- Experimental buffeting alleviation results: (a) -wing tip displacement with the controller on until t = 40s.; (b) Controller signal sent to the actuators to minimize buffeting vibration. 81