Structural Health Monitoring Using Smart Piezoelectric Material

Transcription

1 Structural Health Monitoring Using Smart Piezoelectric Material Kevin K Tseng and Liangsheng Wang Department of Civil and Environmental Engineering, Vanderbilt University Nashville, TN 37235, USA Abstract The application of the electromechanical impedance method to detect the presence of damage and monitor its progression in concrete structures is investigated using finite element analysis. The piezoelectric ceramic (PZT) patch bonded to concrete structures serves as both an actuator and a sensor in high frequency. The various health states of the structure are assessed using the root mean square deviation (RMSD) index. The harmonic response analysis in ANSYS is employed to investigate the correlation of the RMSD index with the location and extent of damage. A series of numerical studies conducted on concrete slabs and concrete beams demonstrate the potential of smart piezoelectric materials for in-situ health monitoring of structural integrity of civil infrastructures. Introduction Structural health monitoring of civil infrastructures is of considerable importance in view of the immense loss of life and property that may result from their failure. The in-situ structural health monitoring technique, that can not only detect an incipient damage, but also make a comprehensive assessment of the structural integrity, is essential for efficient maintenance and management of civil infrastructures. All of the conventional nondestructive evaluation methods have serious limitations and are impractical for in-situ application. Hence, there is a great interest in the development of in-situ health monitoring techniques using piezoelectric materials in the academic and industrial community. Efforts in this field have led to the establishment of three well-known interrogation methods: elastic wave method (Beard and Chang 1997), transfer function method (Chiu et al. 2000), and electromechanical impedance method. Among them, the electromechanical impedance method has received growing attention for in-situ health monitoring due to its distinct advantages. This method utilizes the electromechanical coupling property of piezoelectric materials and detects the damage by monitoring variations of the electric impedance or admittance of the piezoelectric material bonded to structures in high-frequency band. The attractive features of this method include the capability of capturing a wide range of structural damage from small to large scale, no interruption to the service of the structure, ease of implementation into an in-situ automated health monitoring system, and minimum inspection time and effort. The electromechanical impedance method was initially investigated in mechanical and aerospace engineering, and a number of experimental implementations have been performed on complex structures, for instance, an assembled truss structure (Sun et al., 1995), complex precision parts such as gears (Lalande et al., 1996), spot-welded structural joints (Giurgiutiu et al., 1999), a pipe joint (Park et al. 2000), pipelines (Park et al. 2001), jet turbine engine components (Winston et al. 2001), and aircraft metallic specimens (Giurgiutiu et al. 2002). Recently, this method has drawn attention in civil engineering. Successful experimental applications on the concrete and other civil engineering structures (Ayres et al. 1998, Soh et al. 2000, and Tseng and Naidu, 2002) have shown the prospect of the electromechanical impedance method for in-situ health monitoring of civil infrastructures. In contrast to extensive experimental investigations of the impedance method in the literature, little analytical work has been reported to date. This paper presents the application of smart piezoelectric materials for damage detection and characterization in concrete structures by employing finite element analysis. The electromechanical impedance method has been adopted to assess the health states of the structures using the root-mean-square-deviation (RMSD) index. The correlation of the RMSD index with the location and extent of damage is investigated using harmonic response analysis in ANSYS. A series of

2 numerical studies are performed on both concrete slabs and concrete beams. The results illustrate the potential of the electromechanical impedance method for civil infrastructures. Electromechanical Impedance Method An electromechanical impedance model of the dynamic interaction between a PZT patch and its host structure is illustrated in Figure 1. When an alternating electric voltage is applied in the transverse direction, producing an electric field, the PZT patch is considered as a thin bar undergoing axial vibrations in the longitudinal direction in response to the applied electric field. Thus, the excitation of the PZT patch acts as a pair of self-equilibrating axial forces. The electric admittance of the PZT patch is derived in the form (Liang et al. 1994): wplp s 2 ( ) T Z E Y ω = iω ε 33 d31y11 h p Zp + Zs where l p, w p and h p are the length, width and thickness of the PZT patch, respectively, Y = Y (1 +η ) (1) E E i is the Young s modulus at zero electric field, η is the mechanical loss factor, d 31 is the piezoelectric T T constant, ε = ε δ ) is the dielectric constant at zero stress, δ is the dielectric loss factor, ω is the (1 i excitation frequency, Z s is the structural impedance, and Z p is the mechanical impedance of PZT patch defined by Z p E kwphpy 11 = ( iω) tan( kl p ) (2) where E k ω ρ p Y11 = is the wave number, and ρ p denotes the density. Figure 1. Schematic illustration of electromechanical impedance model Eq. (1) indicates that the electric admittance is directly related to the structural impedance. For a given PZT patch bonded to a structure, the structural impedance Z s uniquely determines the electric admittance. The structural impedance is a function of the structural parameters such as stiffness, damping and mass. Therefore, any change in the local structural system due to local damage will lead to changes in the structural impedance, indicating the presence of damage. Since the structural impedance is not easily obtained, we can detect the damage by tracking the electric admittance and comparing it with a pristine measurement. The electric admittance can be conveniently measured by a commercial impedance analyzer.

3 The electric conductance (real part of electric admittance) is usually utilized to predict damage because it is more sensitive to damage than the imaginary part (Sun et al. 1995). The damage can be evaluated quantitatively using the root mean square deviation (RMSD) in conductance signatures with respect to those of healthy state. This non-parametric damage index is defined as: RMSD(%) = n 1 ( M i= 1 i n i= 1 ( M 0 2 Mi ) 0 2 i ) 100 (3) 1 0 where M i and M i are the conductance of the damaged and pristine state at the i-th frequency sampling point, respectively. Since it is very tedious to obtain a closed-form expression for the structural impedance, finite element analysis is used to generate the structural impedance in the numerical study. The effect of the PZT patch is represented with a pair of self-equilibrating harmonic forces F( t) = Fe iω t (4) where F is the amplitude of the harmonic force. The structural impedance at the location of the PZT patch is defined as the applied force divided by its response velocity, v(t): F( t) Z s = (5) v( t) In response to the harmonic exciting force, the response displacement of the PZT patch can be assumed as ξ ( t) = ξ e iω t (6) where ξ is the amplitude of the response displacement. The response velocity of the PZT patch can be written as dξ i t v = = ω iωξ e (7) dt The structural impedance at excitation frequency ω can thus be expressed as Z = s F iωξ (8) If a pair of known harmonic forces is applied to the terminals of the PZT patch simulating its excitation, the response displacement can be obtained from harmonic response analysis using ANSYS. Once the structural impedance is determined, the electric admittance can be calculated by Eq. (1). The RMSD index with respect to the pristine state signatures can then be calculated from Eq. (3). The material properties of the PZT used are as follows: Young s modulus = N/m 2 ; density = 7800 kg/m 3 ; Poisson ratio = 0.3; piezoelectric constant = m/v; dielectric constant = F/m; mechanical loss factor = 0.005; and dielectric loss factor = The material properties of the concrete used are: Young s modulus = 2.25 x N/m 2 ; density = 2400 kg/m 3 ; Poisson ratio = 0.21; mass damping factor = 0.001; stiffness damping factor = Damage Detection in Concrete Slab Using ANSYS A schematic diagram of a concrete slab specimen is illustrated in Figure - Concrete slab with a bonded PZT patch, having the dimensions of mm. The PZT patch bonded on the center of the top of the

4 slab is mm with 0.2 mm in thickness and is excited over frequencies ranging from 40 khz to 60 khz at the interval of 100 HZ. For simplicity, only a pair of horizontal harmonic forces is applied at the terminals of the PZT patch (Points A and B) simulating its excitation. Two types of damage, void and crack, with different models were used to show the applicability of the electromechanical impedance method. Figure 2. Concrete slab with a bonded PZT patch Void detection Figure 3 shows the FE model of a concrete slab with a void. Taking advantage of the symmetry, the FE model of the concrete slab was created with one symmetrical half using SOLID 42 elements in ANSYS 5.7. To ensure the accuracy, finer mesh was created in the region surrounding the PZT patch and the void. Figure 3. FE model of symmetrical half of a concrete slab with a void

5 To study the correlation of the RMSD index with the void size, seven different FE models were created with the diameter of void equal to 0, 3.0, 5.0, 10.0, 15.0, 20.0, 25.0 mm, respectively, while the depth of voids was fixed to 20 mm. Figure 4 shows the variation of the RMSD index with the void size. It is evident that the RMSD index increases as the void size increases. Figure 4. RMSD index with increasing void size To investigate the correlation of the RMSD index with the void depth, two systematic sets of FE models were created with a void, 5.0 mm and 20.0 mm in diameter, respectively, placed at an increasing depth away from the PZT patch. Figure 5 presents the deviation of the RMSD index with the void depth. It is observed that the RMSD index decreases as the void depth increases. It is also noted that an increase of depth less than 20 mm leads to a significant decrease of the RMSD index, while RMSD index is relatively unchanged as the depth goes over 20.0 mm. Figure 5. RMSD index with increasing void depth

6 Crack identification Figure 6 shows the FE model of a concrete slab with a crack. Due to symmetry, only half of the problem domain was created using SOLID 82 elements, SOLID 62 elements, and singular elements in ANSYS 5.7. Figure 6. FE model of symmetrical half of a concrete slab with a crack Six FE models with a crack placed at 20 mm from the PZT patch and having length equal to 0, 5.0, 10.0, 15.0, 20.0, and 25.0 mm, respectively, were created to study the correlation of the RMSD index with the crack length. Figure 7 shows the variation of the RMSD index with the crack length. It is noted that increasing crack length in the vicinity of PZT patch induces an increase of the RMSD index, and the RMSD index does not show appreciable change when the crack grows past the PZT patch. Figure 7. RMSD index with increasing crack length

7 Two systematic sets of FE models with a crack, 5.0 mm and 10.0 mm in length, respectively, placed at an increasing depth away from the PZT patch were used to investigate the correlation of the RMSD index with the crack depth. The variation of RMSD index with the crack depth, as shown in Figure 8, indicates that the RMSD index decreases with the increase of crack depth, and deceases greatly while the crack depth is less than 20.0 mm. Figure 8. RMSD with increasing crack depth Damage Detection in Concrete Beam Using ANSYS Figure 9 illustrates schematically the configuration of the considered concrete beam specimen with the dimensions of mm. Three sets of numerical investigations are conducted: progressive damages on the surface, progressive damages in the depth, and correlation of RMSD index with the damage location. The PZT patch bonded on the concrete beam has the dimensions of mm and is excited over frequencies ranging from 20 khz to 25 khz at the interval of 25 HZ. Figure 9. Schematic of a concrete beam specimen Progressive damages on surface of concrete beam A PZT patch is bonded on the upper surface 40 mm from the left end. Six across-width cracks of 5 mm in depth are simulated sequentially on the upper surface of the beam, located at 300 mm, 250 mm, 200 mm, 150 mm, 100 mm and 50 mm away from the PZT patch, respectively. A FE model is created in ANSYS 6.1 using SOLID 45 element for each damaged state. Due to symmetry, the model of the specimen with one symmetrical half is created with finer mesh in the region surrounding the PZT patch and the cracks. The mesh for the damaged state with four cracks is illustrated in Figure 10. In this analysis, we only consider the x-direction excitation of the PZT patch, and a pair of x-direction harmonic forces is applied at the terminals of the PZT patch.

8 Figure 10. FE model of symmetrical half of a concrete beam with progressive damages on surface The variation of the RMSD index when damage approaches the PZT patch is shown in Figure 11. The farthest damage, located at 300 mm away from the PZT patch, produces an identifiable RMSD value of 1.4. It is also evident from the trend line that the RMSD index increases as the damage approaches the PZT patch. Figure 11. RMSD index with progressive damages approaching PZT on surface

9 Progressive Damages in Depth of Concrete Beam A PZT patch bonded on the right tip end is located 10 mm from the top edge. Seven sequential simulated cracks are located at 360, 310, 260, 210, 160,110, and 60 mm away from the PZT patch, respectively. These across-width cracks have 5 mm in depth. The FE mesh of a symmetrical half of the problem domain for the damaged state with five cracks is shown in Figure 12. Only the y-direction excitation of the PZT patch is considered, and a pair of y-direction forces is applied at the driving points of the PZT patch. Figure 12. FE model of symmetrical half a concrete beam with progressive damages in depth The plot of the variation of the RMSD index versus the crack depth is presented in Figure 13. It can be observed that the farthest damage located at 360 mm away from the PZT patch, is clearly identified, with a noticeable RMSD value of 2.6. It is apparent from the trend line in the figure that the RMSD index increases as the damage extent increases.

10 Figure 13. RMSD index with progressive damages approaching PZT in depth Correlation of RMSD index with damage location A PZT patch is bonded on the upper surface at 40 mm from the left end. An across-width crack of 5 mm in depth is simulated on the upper surface with increasing distance from the PZT patch, 50, 100, 150, 200, 250, 300 mm, respectively. A total of six FE models are created using ANSYS. A pair of x-direction forces is applied at the driving points of the PZT patch to simulate the x-direction response of the PZT patch. The deviation of the RMSD index with the crack distance from the PZT patch is shown in Figure-Variation of RMSD index with distance of single damage. From the linear regression line in the figure it is observed that the RMSD index decreases as the distance of damage from the PZT patch increases. Figure 14. Variation of RMSD index with distance of single damage

11 Conclusion A series of numerical studies have been presented to investigate the application of smart piezoelectric materials for damage detection and characterization in concrete structures using ANSYS. The electromechanical impedance technique has been adopted to assess the various health states of the structures based on global spectrum changes measured by a root mean square deviation (RMSD) on electric conductance over a range of discrete frequencies. Correlation of RMSD index with the location and extent of damage is analytically investigated. The numerical investigations present a strong prospect of employing smart piezoelectric materials in estimating the location and extent of a damage. References 1) Ayers J W, Lalande F, Chaudhry Z and Rogers C A 1998 Qualitative impedance-based health monitoring of civil infrastructures Smat Mater. Struct ) Beard S and Chang F K 1997 Active damage detection in filament wound composite tubes using built-in sensors and actuators J. Intell. Mater. Sys. Struct ) Chiu W K, Galea S C, Koss L L and Rajic N 2000 Damage Detection in Bonded Repairs Using Piezoceramics Smart Mater. Struct ) Giurgiutiu V, Reynolds A and Rogers C A 1999 Experimental investigation of E/M impedance health monitoring for spot-welded structural joints J. Intell. Mater. Syst. Struct ) Giurgiutiu V, Zagrai A and Bao J 2002 Embedded active sensors for in-situ structural health monitoring of thin-wall structures Journal of Pressure Vessel Technology ) Lalande F, Rogers C A, Childs B and Chaudhry Z 1996 High-frequency impedance analysis for NDE of complex precision parts SPIE Symposium on Smart Structures and Materials (San Diego, CA). 7) Liang C, Sun F P and Roger C A 1994 Coupled Electromechanical Analysis of Adaptive Materials-determination of the Power Consumption and System Energy Transfer J. Intell. Mater. Syst. Struct ) Park G, Cudney H H and Inman D J 2000 Impedance-based structural health monitoring of civil structural components Journal of infrastructure systems ) Park G, Cudney H H and Inman D J 2001 Feasibility of using impedance-based damage assessment for pipeline structures Earthquake Engng Struct. Dyn ) Soh C K, Tseng K K-H, Gupta A and Bhalla S 2000 Performance of smart piezoceramic transducers in health monitoring of RC bridges, Smart Mater. Struct ) Sun F P, Chaudhry Z, Liang C and Rogers C A 1995 Truss structure integrity identification using PZT sensor-actuator J. Intell. Mater. Syst. Struct ) Tseng K K-H and Naidu A S K 2002 Non-parametric damage detection and characterization using smart piezoceramic material Smart Mater. Struct ) Winston H A, Sun F and Annigeri B S 2001 Structural health monitoring with piezoelectric active sensors Journal of Engineering for Gas Turbines and Power