Lamb Wave-Based Structural Health Monitoring on ... - Semantic Scholar

materials Article

Lamb Wave-Based Structural Health Monitoring on Composite Bolted Joints under Tensile Load Bin Yang 1 , Fu-Zhen Xuan 1, *, Yanxun Xiang 1 , Dan Li 2 , Wujun Zhu 1 , Xiaojun Tang 3 , Jichao Xu 1 , Kang Yang 1 and Chengqiang Luo 1 1

2 3

*

School of Mechanical and Power Engineering, East China University of Science and Technology, Shanghai 200237, China; [email protected] (B.Y.); [email protected] (Y.X.); [email protected] (W.Z.); [email protected] (J.X.); [email protected] (K.Y.); [email protected] (C.L.) The Department of Electronic Engineering, Fudan University, Shanghai 200433, China; [email protected] Beijing Spacecrafts, China Academy of Space Technology, Beijing 100094, China; [email protected] Correspondence: [email protected]; Tel.: +86-21-64251623

Received: 8 May 2017; Accepted: 11 June 2017; Published: 14 June 2017

Abstract: Online and offline monitoring of composite bolted joints under tensile load were investigated using piezoelectric transducers. The relationships between Lamb wave signals, pre-tightening force, the applied tensile load, as well as the failure modes were investigated. Results indicated that S0/A0 wave amplitudes decrease with the increasing of load. Relationships between damage features and S0/A0 mode were built based on the finite element (FE) simulation and experimental results. The possibility of application of Lamb wave-based structure health monitoring in bolted joint-like composite structures was thus achieved. Keywords: Lamb wave propagation; composite bolted joints; structural health monitoring; FE simulation

1. Introduction The need for stronger and lighter structures in the fields of aerospace, ship engineering, and the automotive industry has driven the use of composite structures. Even though the structure length can be as long as a hundred meters, the maximum dimension of an integrative composite structure is limited into approximately 20 m in length by the manufacturing process, such as the commonly used resin transfer molding (RTM) method [1,2]. Therefore, bolted joints between composite components are necessary when manufacturing large structures [3–5]. However, under real service conditions, damages, e.g., debonding and delamination, in the composite bolted joints are frequently initiated from the fastener holes, which further degenerate structural reliability [6]. Therefore, investigations on the development and optimization of quantitative damage detection and characterization in complex bolted joints are essential. Structural health monitoring (SHM) is defined as nondestructive and continuous monitoring of structures. SHM can identify and monitor the developing defects before their degeneration into a failure [7,8]. In recent years, guided wave-based technology has been shown to be an effective SHM method, attributed to its high sensitivity and versatility [9]. Guided wave can be generated by piezoelectric transducers on the studied structure. This transducer is very light, thin, and can be used as actuator and sensor, respectively [10]. Moreover, guided waves are sensitive to small-scale damages and could propagate over long distance [11]. The Lamb wave is one of the guided wave modalities. The Lamb wave propagates by particle motions between the two surfaces in a thin plate-like medium [12]. Using Lamb wave-based SHM systems, there is no need to scan the entire object under consideration, and all of the data can be acquired from a single probe position. Even though detection and localization Materials 2017, 10, 652; doi:10.3390/ma10060652

www.mdpi.com/journal/materials

Materials 2017, 10, 652

2 of 17

of defects in simple thin-walled composite laminates have been studied extensively [13–15], research on the inspection of complex composite assemblies using Lamb wave-based systems is still insufficient. From the mechanical point of view, most researchers are concerned about the geometrical influence on the bolts of the composite in different composite types with various fiber stacking sequences [16–20]. Researchers claim that the failure modes of composite bolted joints are strongly dependent on the base materials and the designed joint styles. For instance, Zhou et al. [16] investigated the geometric parameter influence on the failure mode of composite single lap bolted joints. The failure modes of glass fiber-reinforced plastic bolted laminates were also extensively studied by Feo et al. [17–19]. Ahmad et al. [20] was concerned with the fracture characteristics in composite double lap bolted joints, and found that the joint finally failed by net-tension damage. Meanwhile, researchers in the mentioned references also found that the structural integrity of the composite joints was threatened severely by the harsh environment during operating life. The joints may result in sharp failure with little advanced warning [21]. Therefore, it is necessary to develop SHM techniques that focus on continuous/real-time monitoring of composite bolted joints. Recently, guided wave technology has been successfully employed for damage detection in bonded joints [22–27]. Mohammad et al. [22] evaluated the integrity of a composite skin-stringer joint using the scattering behavior of Lamb waves. Baiyang et al. [23] used ultrasonic guided waves to investigate the adhesive bonds between composite laminates. Dulip et al. [24] developed a wavelet spectral finite element model for studying transient dynamics and wave propagation features in adhesively bonded composite joints. Guided wave propagation characteristics in bonded composite structures were also investigated by Deng et al. [25]. References devoted to using guided waves to detect defects in the other composite joint types such as T-joint [26] and L-joint [27] were also found. However, very few references concerned the application of Lamb wave-based SHM technology to monitor the damages in composite bolted joints. The major emphasis of this paper was placed on the inspection of composite bolted joints based on the Lamb wave propagation phenomenon. The bolted joints of interest were composed of two cases with different joint length-to-width ratios. The captured Lamb wave signals in the joints before, during and after tensile test were investigated. Lamb wave structures in the researched joints were analyzed, and the relationships between Lamb wave signals, pre-tightening force, the applied tensile load, as well as the failure modes were discussed. Lamb wave propagation behavior in the joints especially in the overlapped jointed region was analyzed using a 3D finite element simulation. All the input parameters in the simulation were determined experimentally by fundamental mechanical tests. 2. Experimental Details 2.1. Materials and Processing The polymer used in the composites as the matrix was vinyl ester resin. The resin can be cured at room temperature in the presence of a hardening agent and an accelerating agent. The hardening agent was methyl ethyl ketone peroxide (MEKP), and the accelerating agent was dimethylaniline. The resin was mixed with hardening agent and accelerating agent at mass ratio of 1:2:0.5. The reinforced material was plain woven glass fabric cloth (WGF). We adopted a vacuum-assisted resin injection (VARI) process to manufacture the WGF/epoxy composites in the bolted joints. Figure 1 shows the schematic diagram of the VARI process. During the manufacturing process, we used a glass plate at the bottom of the VARI system as a holder. After coating the release cloth on the holder, we placed four layers of woven glass fabric. The stacking sequence of the plate was [90◦ /0◦ ]4 . After the system was covered with peel and silk ply, the system was closed by a vacuum bag and sealant tape. We vacuumed the system at 600 mbr at room temperature for 10 h to ensure complete vacuum conditions. To ensure the resin flowing uniformly, a delivery pipe was fixed at the entrance. After resin injection and curing for 24 h, the composites were demolded. The final dimension of the manufactured WGF/epoxy composite laminates was 800 × 800 × 2 mm3 , and we cut it into the required dimension by a low speed diamond saw blade cutting machine under water protection.

Materials 2017, 10, 652

3 of 17

Materials 2017, 10, 652

3 of 17

Figure 1. Schematic diagram of the vacuum-assisted resin injection (VARI) manufacturing process

Figure 1. Schematic diagram of the vacuum-assisted resin injection (VARI) manufacturing process and and the two studied composite bolted joints with the coordinate system. the two studied composite bolted joints with the coordinate system.

The fastener holes on WGF/epoxy composite laminates were made by an electric drill, and the diameter of theholes holeson were 10 mm. It should be noted that the were drill speed very slow, The fastener WGF/epoxy composite laminates madewas by set an to electric drill,and and the initialofdamage waswere not formed after the drilling operations. Twodrill composite joint cases slow, were and diameter the holes 10 mm. It should be noted that the speed bolted was set to very Case A formed were composite joints with aoperations. single bolt, while joints in Case B had two steel were initialinvestigated: damage was not after the drilling Two the composite bolted joint cases bolts in them. As shown in Figure 1 and listed in Table 1, each case included three categories of joints investigated: Case A were composite joints with a single bolt, while the joints in Case B had two steel with the jointed length (L) to width (W) ratio (L/W) various from 1:1 to 3:1. Meanwhile, the coordinate bolts in them. As shown in Figure 1 and listed in Table 1, each case included three categories of joints system adopted in the following work was given in Figure 1. The total length and the bolt locations with the jointed length (L) to width (W) ratio (L/W) various from 1:1 to 3:1. Meanwhile, the coordinate in the joints were listed in Table 1. system adopted in the following work was given in Figure 1. The total length and the bolt locations in the joints Table were1.listed in Table 1. geometric dimension of the two studied composite bolted joints in the Structural style and experiment.

Table 1. Structural style and geometric dimension of the two studied composite bolted joints in Parameter Case A Case B the experiment. W 40 40 L 40 80 Parameter Case L/W 1:1 2:1 A total 450 410 LW 40 40 L Bolt location (L/2,40 W/2) (L/2,80 W/2) L/W 1:1 2:1 Ltotal 450 410 2.2. Fundamental Mechanical Tests Bolt location (L/2, W/2) (L/2, W/2)

40 120 3:1 380 40 120 (L/2, W/2) 3:1 380 (L/2, W/2)

60 60 1:1 450 60 60W/4) (L/2, 1:1 450 (L/2, W/4)

60 90 Case 3:2B 410 60 90 W/4) (L/2, 3:2 410 (L/2, W/4)

60 120 2:1 380 60 120W/4) (L/2, 2:1 380 (L/2, W/4)

In order to determine the stiffness and strength parameters of WGF/epoxy composites in the bolts, fundamental mechanical 2.2. Fundamental Mechanical Tests tests that included 90°, 0° and ±45° tensile and compression tests were accomplished using an INSTRON-4505 servo-electric testing machine (Instron, Grove City, PA, USA).

In to determine the stiffness and strength parameters WGF/epoxy composites Theorder nominal maximum force of the adopted test machine was 500of kN. The dimensions of the testin the ◦ ◦ ◦ bolts,specimens fundamental included Society 90 , 0 for and ±45 Materials tensile and compression were mechanical in accordancetests withthat the American Testing (ASTM) standard. tests to evaluate the performance between adjacent glass fabric cloth layers, we tested were Specially, accomplished using aninterfacial INSTRON-4505 servo-electric testing machine (Instron, Grove City, PA, interlaminate performance of the composites double cantilever (DCB), endof the USA).the The nominal maximum force of the adoptedbytest machine was beam 500 kN. Thethree-point dimensions notched flexural (3ENF), and Mode III fracture toughness tests, respectively. It should be noted that a test specimens were in accordance with the American Society for Testing Materials (ASTM) standard. pre-crack with a length of 25 mm was laid in the middle thickness of each specimen. The used Specially, to evaluate the interfacial performance between adjacent glass fabric cloth layers, we tested mechanical testing standards, specimen’s dimension, and the calibrating length of different tests were the interlaminate performance of the composites by double cantilever beam (DCB), three-point end listed in Table 2. All of the mentioned mechanical tests were performed at room temperature. The notched flexural (3ENF), and Mode III fracture toughness tests, respectively. It should be noted that cross-head speed was set as 2 mm/min, and five specimens were tested per type to calculate their a pre-crack average with value.a length of 25 mm was laid in the middle thickness of each specimen. The used mechanical testing standards, specimen’s dimension, and the calibrating length of different tests were listed in Table 2. All of the mentioned mechanical tests were performed at room temperature. The cross-head speed was set as 2 mm/min, and five specimens were tested per type to calculate their average value.

Materials 2017, 652 Materials 2017, 10,10, 652

4 of417 of 17

Figure 2 demonstrates the average stress-strain and load-displacement curves of the WGF/epoxy Figure 2 demonstrates the average stress-strain andInload-displacement curves thetensile WGF/epoxy composite laminates in different mechanical tests. detail, Figure 2a–d are of the and composite laminates inalong different tests. In detail, 2a–d are the tensile andthree-point compression compression curves 90°,mechanical 0°, and ±45° of the fibers, Figure respectively. Figure 2e is the ◦ , 0◦ , while bending test90 curve, are the respectively. load-displacement curves in DCB, bending 3ENF andtest curves along and ±Figure 45◦ of2f–h the fibers, Figure 2e isobtained the three-point the Mode test, 2f–h respectively. The elasticity, strength, fracture together curve, while III Figure are the load-displacement curvesand obtained in toughness DCB, 3ENFparameters and the Mode III test, with their standard deviations of theand WGF/epoxy composite laminates determined from their Figure 2 are respectively. The elasticity, strength, fracture toughness parameters together with standard listed in Tables 3–5. In these tables, the longitudinal and transverse modulus E11/Ein 22, Tables and the3–5. deviations of the WGF/epoxy composite laminates determined fromelastic Figure 2 are listed shear modulus G 12 were calculated using the initial tensile curve slope in Figure 2a–d. The In these tables, the longitudinal and transverse elastic modulus E11 /E22 , and the shear modulus G12 longitudinal/transverse tensiletensile strength (Xt/Yslope t), compression strength (Xc/Yc), and the shear strength were calculated using the initial curve in Figure 2a–d. The longitudinal/transverse tensile S were the peak stresses in Figure 2a–d that were determined bystrength the maximum by the in strength (Xt /Yt ), compression strength (Xc /Yc ), and the shear S wereload thedivided peak stresses cross-section area of the longitudinal, transverse, and 45° tensile specimens, respectively. The elastic Figure 2a–d that were determined by the maximum load divided by the cross-section area of the modulus E11, E22 and G can be calculated by the following equation: longitudinal, transverse, and 45◦ tensile specimens, respectively. The elastic modulus E11 , E22 and G can be calculated by the following equation: S 3m (1) E 3 3m 4Sbh (1) E= 4bhm3 is the slope of the load-displacement curve, where E is the modulus, and S is the support span. whileE bisand are the width respectively. The following formulas were adopted where thehmodulus, andand S isthickness, the support span. m is the slope ofthree the load-displacement curve, to calculate the Mode I, Mode II, and Mode III interlaminar fracture toughness (ILFT) of the WGF/epoxy while b and h are the width and thickness, respectively. The following three formulas were adopted to composite laminates according to Figure 2f–h: calculate the Mode I, Mode II, and Mode III interlaminar fracture toughness (ILFT) of the WGF/epoxy 3P  composite laminates according to Figure 2f–h: GⅠc  c c (2)

2ba

3Pc δc GIc = 2 9a2ba 0 Pc c

GⅡc  GIIc =

(2)

3 c3δac 03 ) 2b(9al032 P 8 3 3 2b( l + 3a 3 ) 8

(3)

(3)

0

S G III c =  ∆S G IIIc 2b a 2b∆a

(4) (4)

where, GIcG , G and GIIIc . are interlaminar fracture toughness; a0 and a are the pre-crack length GIIIctheare where, ⅠcIIc, GⅡc and the interlaminar fracture toughness; a0 and a are the pre-crack and crack propagating length, respectively. Pc and δc are the load and displacement when the crack length and crack propagating length, respectively. Pc and δc are the load and displacement when the propagates to a specific length ∆a, while b and l are the width and length of the specimens, respectively. crack propagates to a specific length Δa, while b and l are the width and length of the specimens, ∆S is the area covered by Pc and δc , as indicated in Figure 2h. The minor deviations of experimental respectively. ΔS is the area covered by Pc and δc, as indicated in Figure 2h. The minor deviations of results in Tables 3–5 may be caused by uncertainty factors in the manufacturing process of the experimental results in Tables 3–5 may be caused by uncertainty factors in the manufacturing process anisotropic composite materials [28,29]. Specially, in Table 5, the maximum loads in the three tests are of the anisotropic composite materials [28,29]. Specially, in Table 5, the maximum loads in the three 112.5 N, 1628.4 N, and 512.3 N, with corresponding failure displacements of 5.45 mm, 8.78 mm, and tests are 112.5 N, 1628.4 N, and 512.3 N, with corresponding failure displacements of 5.45 mm, 8.78 mm, 2 10.67 The ILFT WGF/epoxy specimens is 1.65, 2.71, and 1.22 kJ/m andmm, 10.67respectively. mm, respectively. The of ILFT of WGF/epoxy specimens is 1.65, 2.71, and 1.22 kJ/m,2,which whichcan can be used to evaluate the interface performance of WGF/epoxy composite laminates in the following work. be used to evaluate the interface performance of WGF/epoxy composite laminates in the following work.

(a)

(b)

Figure 2. Cont.

Materials 2017, 10, 652

5 of 17

Materials 2017, 10, 652

5 of 17

(c)

(d) 120

*

Mode I fracture toughness

100

Load/N

80 Initial cracking

60 40 20 0

10

15

Displacement/mm

20

25

(f) 600

Mode II fracture toughness

1600

Mode III fracture toughness

500

1400 1200

400

1000

Load/N

Load/N

5

(e)

1800

Complete cracking

800

Initial cracking

Flexural damage

600

300 200

400

ΔS

100

200 0

0

0

2

4

6

8

Displacement/mm

(g)

10

12

14

0

0

3

6

9

12

15

18

Displacement/mm

21

24

27

(h)

Figure 2. Mechanical performance of woven glass (WG)/epoxy composite laminates obtained in different Figure 2. Mechanical performance of woven glass (WG)/epoxy composite laminates obtained in experimental tests. (a): 90°/0° tensile; (b): 45° tensile; (c): 90°/0° compression; (d): 45° compression; different experimental tests. (a): 90◦ /0◦ tensile; (b): 45◦ tensile; (c): 90◦ /0◦ compression; (d): 45◦ (e): Three-point-bending test; (f): Mode-I fracture toughness test; (g): Mode-II fracture toughness test; compression; (e): Three-point-bending test; (f): Mode-I fracture toughness test; (g): Mode-II fracture (h): Mode-III fracture toughness test. toughness test; (h): Mode-III fracture toughness test. Table 2. Test standard and dimension of woven glass fabric (WGF)/epoxy in different experiments. Table 2. Test standard and dimension of woven glass fabric (WGF)/epoxy in different experiments. Calibrating Length Dimension Test Method ASTM Standard (mm) (L × WDimension × T, mm3) Calibrating Test Method ASTM Standard 3 90°/0° tensile D3039/D3039M-08 220 × 20 × 2.5 140 (L × W × T, mm ) Length (mm) ±45° tensile D3518/D3518M-13 180 × 20 × 2.5 100 90◦ /0◦ tensile D3039/D3039M-08 220 × 20 × 2.5 140 ±45°/90°/0° D3410/D3410M-95 160180 × 20 80 100 ±45◦ compression tensile D3518/D3518M-13 × ×202.5× 2.5 ◦ ◦ ◦ DCB D5528-13 220 × 25 × 2.5 25 (precrack) ±45 /90 /0 compression D3410/D3410M-95 160 × 20 × 2.5 80 3ENF D7905/D7905M-14 100220 × 25 DCB D5528-13 ××252.5× 2.5 2580(precrack) Mode E1922-04 100100 × 25 25 (precrack) 3ENFIII D7905/D7905M-14 ××252.5× 2.5 80 Mode III Society for Testing Materials; E1922-04 DCB: double cantilever 100 × 25 ×beam; 2.5 3ENF: three-point 25 (precrack) ASTM: American end ASTM: American Society for Testing Materials; DCB: double cantilever beam; 3ENF: three-point end notched notched flexural. flexural.

Materials 2017, 10, 652

6 of 17

Table 3. Elastic parameters obtained in the mechanical test. Materials 2017, 10, 652

6 of 17

E11 /GPa

E22 /GPa

E33 /GPa

11.8 ± 0.23

11.8 ± 0.25

0.58 ± 0.023. Table

E11/GPa 11.8 ± 0.23

G12 /GPa

G13 /GPa

G23 /GPa

µ12

µ13

ρ/(g/cm3 )

µ23

4.82 ± 0.08 4.82 ± 0.08 4.82 ± in 0.08the 0.05 ± 0.01 0.24 ± 0.01 Elastic parameters obtained mechanical test.

0.23 ± 0.01

E22/GPa E33/GPa G12/GPa G13/GPa G23/GPa µ12 µ13 µ23 Strength the mechanical 11.8 ± 0.25 Table 0.58 ±4.0.02 4.82 ±parameters 0.08 4.82 ± obtained 0.08 4.82in ± 0.08 0.05 ± 0.01 test 0.24(MPa). ± 0.01 0.23 ± 0.01

XT

XC

Y

Y

Z

Z

S

S

T T 13 C Table 4. Strength parameters obtained in the Cmechanical12test (MPa).

485.05 ± 59.94 ± XT 12.54 XC2.37 485.05 ± 59.94 ± 12.54 2.37

488.91 ± YT 15.29 488.91 ± 15.29

59.58 ± YC 1.27 59.58 ± 1.27

52.14 ± ZT 0.98 52.14 ± 0.98

1.65

ρ/(g/cm3) 1.65

S23

180.24 ± 108.4 ± 108.4 ± 108.4 ± 1.89ZC 5.34 S12 5.34 S13 5.34 S23 180.24 ± 108.4 ± 108.4 ± 108.4 ± 1.89 5.34 5.34 5.34

Table 5. Interlaminate parameters of WGF/epoxy composites calculated from fracture toughness tests.

Table 5. Interlaminate parameters of WGF/epoxy composites calculated from fracture toughness tests. Test Maximum Failure Strain Energy Release 2) Method Load/N Displacement/mm Rate/(kJ/m Failure Strain Energy Release Test Method

Mode-I Mode-I Mode-II Mode-II Mode-III

Mode-III

Maximum Load/N

112.5 ± 3.24 112.5 ± 3.24 1628.4 ± 20.05 ± 20.05 512.31628.4 ± 4.34 512.3 ± 4.34

5.45Displacement/mm ± 0.05 5.45 ± 0.05 8.78 ± 0.32 10.67 ± 8.78 1.12± 0.32 10.67 ± 1.12

) 1.65 ±Rate/(kJ/m 0.05 1.65 ± 0.05 2.71 ± 0.03 2.71 ± 0.03 1.22 ± 0.01 2

1.22 ± 0.01

2.3. 2.3. Lamb Lamb Wave Wave Inspection Inspection System System Figure 3 shows the experimental setup forfor Lamb wave propagation in theincomposite boltedbolted joints Figure 3 shows the experimental setup Lamb wave propagation the composite in CaseinACase and Case B.Case In theB.setup, single channel function generator (AFG3021C, Tektronix, joints A and In thea setup, a single arbitrary channel arbitrary function generator (AFG3021C, INC., Beaverton, OR, USA) was adopted to excite the transmitting transducers. The actuating signal Tektronix, INC., Beaverton, OR, USA) was adopted to excite the transmitting transducers. The actuating consisted of a 3.5-cycle sinusoidal tone burst. The signal had a frequency of 210 kHz and was modulated signal consisted of a 3.5-cycle sinusoidal tone burst. The signal had a frequency of 210 kHz and was by a Hanning The actuating signal was then divided into divided two groups: one group was as modulated bywindow. a Hanning window. The actuating signal was then into two groups: oneused group the input wave into the composite bolted joints after being amplified 20 times by a linear amplifier was used as the input wave into the composite bolted joints after being amplified 20 times by a linear (Mode EPA-104, Systems, Cambridge, MA, USA), while the while other group wasgroup collected amplifier (Mode Piezo EPA-104, PiezoINC., Systems, INC., Cambridge, MA, USA), the other was directly by the oscilloscope (MDO3021, Tektronix, INC., Beaverton, OR, USA). The signals were then collected directly by the oscilloscope (MDO3021, Tektronix, INC., Beaverton, OR, USA). The signals transferred to a PC for to data processing a general purpose interface businterface (GPIB) interface. It were then transferred a PC for data through processing through a general purpose bus (GPIB) should be noted that the transmitting transducers adopted in the experiments interface. It should be noted that the transmitting transducers adopted inwere the piezoelectric experiments (PZT) were wafers, and they had a circular shape with a diameter of 10 mm, and a thickness of and 1 mm. All PZT piezoelectric (PZT) wafers, and they had a circular shape with a diameter of 10 mm, a thickness transducers were bonded on the were surface of the composite boltedofjoints using cyanoacrylate adhesive. of 1 mm. All PZT transducers bonded on the surface the composite bolted joints using Two PZT wafers were tied symmetrically on both of the joint on specimen with cyanoacrylate adhesive. Two PZT wafers were tiedends symmetrically both ends ofan theactuator-sensor joint specimen distance of 215 mm, thusdistance the signals hadmm, to travel thehad jointtoregion. with an actuator-sensor of 215 thus through the signals travel through the joint region.

Figure 3. Experimental setup of the Lamb wave-based Structural Structural health health monitoring monitoring (SHM) (SHM) system. system.

To investigate the pre-tightening force effect of the steel bolt on the wave propagation behavior To investigate the pre-tightening effect of 20 thetosteel bolt on on the the joints wave using propagation behavior in the composite joints, we applied theforce torque from 30 N·m a digital display in the composite joints, we applied the torque from 20 to 30 N · m on the joints using a digital torque wrench. The wave signals in the two cases with different torques were collected by thedisplay system torque wrench. wavewere signals in the two with different torques wereofcollected by the in in Figure 3. TheThe signals recorded withcases a torque incremental interval 1 N·m. All thesystem obtained data were compared and analyzed in terms of wave amplitude. To evaluate the relationship between the waveform and the applied load on the composite bolted joints, we performed tensile tests on the joint specimens. During the tensile test, the load-displacement curves were collected, and the corresponding Lamb waves were captured. The load speed was also set as 2 mm/min at room temperature.

Materials 2017, 10, 652

7 of 17

Figure 3. The signals were recorded with a torque incremental interval of 1 N·m. All the obtained data were compared and analyzed in terms of wave amplitude. To evaluate the relationship between the waveform and the applied load on the composite bolted joints, we performed tensile tests on the joint specimens. During the tensile test, the load-displacement curves were collected, and the corresponding Lamb waves were captured. The load speed was also set as 2 mm/min at room temperature. 3. Fundamental Theory and Dispersion Feature of Lamb Wave in Composite Laminates Materials 10, has 652 Lamb 2017, wave

7 of 17 has a nature of dispersion and a multi-mode phenomenon, and it generally symmetric (S) and antisymmetric (A) modes during propagation [8]. In an isotropic medium, the 3. Fundamental Theory and Dispersion Feature of Lamb Wave in Composite Laminates constituent properties, geometry, wave direction, and frequency could affect the Lamb wave modes [11]. Lamb wave hasmodes a nature of dispersion and athey multi-mode and it generally Meanwhile, the wave may change when meet a phenomenon, structural boundary and/orhas defects. symmetric (S) and antisymmetric (A) modes during propagation [8]. In an isotropic medium, the However, due to anisotropic properties, wave propagation behavior in multi-layered composite constituent properties, geometry, wave direction, and frequency could affect the Lamb wave modes [11]. laminates is very complex. The wave interactions depend upon not only the individual constituents Meanwhile, the wave modes may change when they meet a structural boundary and/or defects. but also their interface properties. In an N-layered composite laminate, there are six wave modes in total However, due to anisotropic properties, wave propagation behavior in multi-layered composite that could define a single guided in laminated plates two quasi-longitudinal laminates is very complex. Thewave wavemode interactions dependcomposite upon not only the[12]: individual constituents (L), two quasi-shear vertical waves (SV), and two quasi-shear horizontal waves (SH). The Lamb but also their interface properties. In an N-layered composite laminate, there are six wave modes in wave can be generally described its displacement displacement total that could define using a single guided wave field modebyinsatisfying laminatedNavier’s composite plates [12]: equations two quasi-longitudinal quasi-shear vertical waves (SV), and two quasi-shear horizontal waves (SH). within each layer [30],(L), astwo following:

The Lamb wave can be generally described using its displacement field by satisfying Navier’s 2 n displacement equations withinn each asnfollowing:n 2 nlayer [30], n n∂ µ

µ ∇ µ + (λ + µ )∇(∇·µ ) = ρ

∂t2

(5)

2 n n 2 n n n n n   (5)              ( ) ( ) where, ρi and λi , µi are the density, Lame constant and displacement t 2 for the ith layer, respectively.

The selection of the appropriate Lamb wave mode and frequency could deeply affect the wave i where,  i and  ,  i are the density, Lame constant and displacement for the ith layer, respectively. structure analysis processing due to the dispersive nature. This is further complicated during The selection of the appropriate Lamb wave mode and frequency could deeply affect the wave the propagation into anisotropic material due to the complex structural features. To find out the structure analysis processing due to the dispersive nature. This is further complicated during the suitable frequency for effectively monitoring the composite bolted joint, we calculated the dispersion propagation into anisotropic material due to the complex structural features. To find out the suitable characteristics by solving Equation (5) in software 4 depictsthethedispersion phase/group frequency for effectively monitoring theDisperse composite bolted [31]. joint, Figure we calculated velocity-frequency a boundary-free WGF/epoxy composite laminated with a stacking characteristics bycurves solvingof Equation (5) in Disperse software [31]. Figure 4 depicts the phase/group sequence of [90◦ /0◦ ]4 .curves The dispersion curves in Figure 4 clearly displayed thewith symmetric (S) and velocity-frequency of a boundary-free WGF/epoxy composite laminated a stacking sequence of [90°/0°] 4 . The dispersion curves in Figure 4 clearly displayed the symmetric (S) and antisymmetric (A) Lamb wave modes. As highlighted with a dotted rectangle in Figure 4, a less antisymmetric (A) Lamb modes. As highlighted with the a dotted rectangle in Figure 4, a at less dispersive region existed in wave the low frequency range where S0 and A0 modes travelled almost dispersive regionThis existed in the low frequency the S0 and A0 modes at almost constant velocities. frequency is referred to range as thewhere non-dispersion region. Intravelled detail, there were only constant velocities. This frequency is referred to as the non-dispersion region. In detail, there were A0 and S0 wave modes when the excited frequency was less than 400 kHz. In the following work, we only A0 and S0 wave modes when the excited frequency was less than 400 kHz. In the following selected the incident wave with a frequency of 210 kHz to monitor the health status of the composite work, we selected the incident wave with a frequency of 210 kHz to monitor the health status of the bolted joints. This selected could simplifycould the waveform analysis in post-processing since it composite bolted joints.frequency This selected frequency simplify the waveform analysis in postcould minimize the amplitude of unwanted mode signals. processing since it could minimize the amplitude of unwanted mode signals. 4

10

Phase Velocity/(km/s)

S2 A2

A3 A4

S1

S0

Group Velocity/(km/s)

S1

8

S3

A1 6 4 S0 2

S2

3

A1

2

A2 1

S3 A3

A0

A4

A0 0 0.0

0.4

0.8

1.2

Frequency/MHz

1.6

2.0

0 0.0

(a) phase velocity

0.4

0.8

1.2

Frequency/MHz

1.6

2.0

(b) group velocity

Figure 4. Dispersion property of WGF/epoxy composite laminates with a stacking sequence of [90°/0°]4.

Figure 4. Dispersion property of WGF/epoxy composite laminates with a stacking sequence of [90◦ /0◦ ]4 .

4. Experimental Works on SHM of the Composite Bolted Joints Using Lamb Wave 4.1. Lamb Wave in Composite Bolted Joints Before Tensile Test Figure 5a–c depict the typical Lamb wave, A0/S0 amplitude and their ratio tendency in the composite bolted joints, respectively. It should be noted that the tested specimens in Figure 5a are in

Materials 2017, 10, 652

8 of 17

4. Experimental Works on SHM of the Composite Bolted Joints Using Lamb Wave 4.1. Lamb Wave in Composite Bolted Joints Before Tensile Test Figure 5a–c depict the typical Lamb wave, A0/S0 amplitude and their ratio tendency in the composite bolted joints, respectively. It should be noted that the tested specimens in Figure 5a are in the free boundary condition without tensile load. For easy observation, we gave the incident Lamb wave and the reception signals by the actuator and sensor pairs, simultaneously. The received waves were formed by different sources such as direct wave, the inner scatter, and boundary reflection waves. In Figure 5a, the propagation duration of the direct Lamb wave in different cases matched the S0 mode in the dispersion curves in Figure 4. The predicted phase/group velocity was calculated by dividing theMaterials actuator-sensor distance (215 mm) by the arriving time; this was 3700 m/s. The first wave 2017, 10, 652 8 of 17 package was the S0 mode Lamb wave that propagated directly from the actuator to the sensor, and its and thewas reception by According the actuator and pairs, simultaneously. The phase/group received waves velocity of group/phasewave velocity the signals largest. tosensor the dispersion curve, the were formed by different sources such as direct wave, the inner scatter, and boundary reflection the A0 Lambwaves. waveInwas 1500 m/s, thus the wave package with arriving times approximately of 140 µs Figure 5a, the propagation duration of the direct Lamb wave in different cases matched the was the A0 mode Lamb package, marked Figurephase/group 5a. There are normally two wave packages S0 mode in the wave dispersion curves inasFigure 4. Theinpredicted velocity was calculated by actuator-sensor distanceare (215formed mm) by the this scattering was 3700 m/s.effect The first waveas reflection, between A0 dividing and S0the wave modes, which byarriving Lambtime; wave such package was the S0 mode Lamb wave that propagated directly from the actuator to the sensor, and transmissionitsand mode conversion when they meet the discontinuity in the overlapped joint region. group/phase velocity was the largest. According to the dispersion curve, the phase/group velocity As shown inofFigure 5b,c,wave the was amplitude of the mode was thanapproximately that of A0 mode the A0 Lamb 1500 m/s, thus the S0 wave package withsmaller arriving times of 140 µsin both cases. was the modebecause Lamb wave marked Figure 5a. There are normally two wave on the tone This was mainly theA0case thatpackage, S0 andasA0 modein are excited differently, depending packages between A0 and S0 wave modes, which are formed by Lamb wave scattering effect such as burst excitation frequency [32–34]. Furthermore, since it is difficult for in-plane particle motion to reflection, transmission and mode conversion when they meet the discontinuity in the overlapped cross the interface, S0 As mode isinmostly retained in theofplate, partial out-of-plane A0 joint region. shown Figure 5b,c, the amplitude the S0 while mode was smallerenergy than thatof of the A0 mode in both cases. was mainly the case because S0 and A0 the mode are excited modes will leak into the This interface. Another aspect, wethat found that Lamb wave differently, amplitude increased depending on the tone burst excitation frequency [32–34]. Furthermore, since it is difficult for in-plane with increasing of L/W ratio. This phenomenon can be explained as following: In the joints, the particle motion to cross the interface, S0 mode is mostly retained in the plate, while partial energy of discontinuity occurs at the andthethe bolt region. This leads to that a result where the out-of-plane A0 ending modes willedge leak into interface. Another aspect, we found the Lamb wave the incident amplitude increased with increasing of L/W ratio. This phenomenon can be explained as following: Lamb wave will be partially received by the sensor, and partially reflected. Since the overlap area In the joints, the discontinuity occurs at the ending edge and the bolt region. This leads to a result increased, the Lamb wave energy during its propagation in the joints with higher L/W ratio also where the incident Lamb wave will be partially received by the sensor, and partially reflected. Since increased. This mechanism further to the larger wave amplitude. Moreover, the amplitude in the overlap area increased, the leads Lamb wave energy during its propagation in the joints with higher L/W 5b) ratiowas also increased. Thislarger mechanism leads to the wave amplitude. the due to the Case A (Figure relatively thanfurther that of Case B larger (Figure 5c), whichMoreover, was mainly amplitude in Case A (Figure 5b) was relatively larger than that of Case B (Figure 5c), which was increased number of fastener holes that could dissipate the wave energy by reflection. mainly due to the increased number of fastener holes that could dissipate the wave energy by reflection. 0.15

A0

Single bolt 1:1 Envelope

0.10 S0

Amplitude

0.05 0.00

-0.05 -0.10 -0.15

0

100

200

Time/microsecond

300

400

(a) Typical time domain signal

0.4

S0 Mode A0 Mode A0/S0 Ratio

0.20

2.5

3.5

S0 Mode A0 Mode A0/S0 Ratio

3.0

0.15

2.5

1.0 0.1 0.0

0.5

1.0

1.5

2.0

L/W ratio

(b) Case-A

2.5

3.0

0.0

Amplitude

1.5 0.2

Ratio

Amplitude

2.0 0.3

2.0 0.10 1.5

Ratio

3.0

0.5

1.0

0.05

0.5 0.00

1.0

1.2

1.4

1.6

L/W Ratio

1.8

2.0

0.0

(c) Case-B

Figure 5. Waveform characteristics and the amplitude tendency in different composite bolted joints Waveform characteristics and theonamplitude tendency different composite bolted without tensile load (the applied torque the bolts in the joints is 20 in N·m).

Figure 5. without tensile load (the applied torque on the bolts in the joints is 20 N·m).

Also, the interface between two jointed composite plates could affect the wave energy and cause the wave amplitude decline by reflecting and scattering during its propagation. As known, in the

joints

Materials 2017, 10, 652

9 of 17

Also, the interface between two jointed composite plates could affect the wave energy and cause Materials 2017, 10, 652 9 of 17 the wave amplitude decline by reflecting and scattering during its propagation. As known, in the composite between thethe two jointed specimens cancan be evaluated by the composite bolted boltedjoints, joints,the theinterface interfacestate state between two jointed specimens be evaluated by torque on the bolts during the assembling process. In order to evaluate the interface bonding effect the torque on the bolts during the assembling process. In order to evaluate the interface bonding on Lamb wave wave behavior in composite bolted joints,joints, Figure 6 describes the relationship between the effect on Lamb behavior in composite bolted Figure 6 describes the relationship between amplitude of S0ofand waves and the on theon bolts different composite joints cases. the amplitude S0 A0 andmode A0 mode waves andtorque the torque the in bolts in different composite joints As shown in the figure, the amplitude of both the S0 and A0 mode waves showed an approximately cases. As shown in the figure, the amplitude of both the S0 and A0 mode waves showed an decreasing tendency with an increasewith of torque. The wave amplitude in Case A was generally larger approximately decreasing tendency an increase of torque. The wave amplitude in Case A was than that of Case B. As discussed, this was attributed to the wave reflection phenomenon in Case B generally larger than that of Case B. As discussed, this was attributed to the wave reflection being more severe than that of Case A. To make signal analysis processing easier, we applied a torque phenomenon in Case B being more severe than that of Case A. To make signal analysis processing of 20 N·we m on all theabolts, since their S0 on andallA0the wave amplitudes were easier distinguish among all easier, applied torque of 20 N·m bolts, since their S0 and A0towave amplitudes were the complex wave packages in the received signals. easier to distinguish among all the complex wave packages in the received signals. 0.16

S0 Mode/Case-A A0 Mode/Case-A S0 Mode/Case-B A0 Mode/Case-B

0.14

Amplitude

0.12 0.10 0.08 0.06 0.04 0.02 0.00

20

22

24

26

Torque/N.m

28

30

Figure 6. 6. Amplitude Amplitude of of S0 S0 and and A0 A0 mode mode as as the the function function of of the the applied applied torque torque on on the the bolt. bolt. Figure

4.2. Lamb Lamb Waves Waves Inspection Inspection of of the the Composite Composite Bolted Bolted Joints Joints after after Tensile TensileTest Test 4.2. Figure7 7illustrates illustrates representative curves of macroscopic load-deflection (L-D) of the Figure thethe representative curves of macroscopic load-deflection (L-D) of the composite composite bolted joints in tensile test. It can be seen from the experimental results that the tensile bolted joints in tensile test. It can be seen from the experimental results that the tensile load load appeared initially toapproximately be an approximately linear relationship, with plasticwith yielding with appeared initially to be an linear relationship, and withand plastic yielding increased increased displacement. The failure loads were approximately 6300 N and 12,000 N for Case A displacement. The failure loads were approximately 6300 N and 12,000 N for Case A and Caseand B, Case B, respectively. In terms of joint configuration, the potential failure modes in bolted composite respectively. In terms of joint configuration, the potential failure modes in bolted composite joints are joints are comprised of cleavage (C), net-tension (N), shear out (S), and bearing (B) [16,18]. Figure 8 comprised of cleavage (C), net-tension (N), shear out (S), and bearing (B) [16,18]. Figure 8 shows the shows the final damage pattern after the experiments. Since the strength of the steel bolt is much final damage pattern after the experiments. Since the strength of the steel bolt is much higher than that higher than that of the WGF/epoxy composites, composites damages were the main failure modes. of the WGF/epoxy composites, composites damages were the main failure modes. Damages occurred Damages occurred on L/W=1:1 bolted joints are the comprehensive failure modes that comprised of on L/W=1:1 bolted joints are the comprehensive failure modes that comprised of B + S + N, while the B + S + N, while the other specimens appeared for the failure mode as B + N. It is worth mentioning other specimens appeared for the failure mode as B + N. It is worth mentioning that during the test, that during the test, the occurring sequence of failure mode in the composite plate was B > S > N. In the occurring sequence of failure mode in the composite plate was B > S > N. In the tensile test, the the tensile test, the tensile stress in the joints was mainly supported by the shear load from the used tensile stress in the joints was mainly supported by the shear load from the used bolts, which further bolts, which further limited the tensile peak load in Figure 7. This stress firstly caused bearing failure (B) limited the tensile peak load in Figure 7. This stress firstly caused bearing failure (B) around the bolt around the bolt hole, and then shear damage appeared. It should be noted that this process was hole, and then shear damage appeared. It should be noted that this process was accompanied by accompanied by other failures such as delamination and fiber/matrix cracking. These failure modes other failures such as delamination and fiber/matrix cracking. These failure modes in the composite in the composite panel further led to the final failure mode (N) on the specimens. Since fiber breaking panel further led to the final failure mode (N) on the specimens. Since fiber breaking in the composite in the composite specimens progressively occurred, it resulted in a ‘load plateau’ in the load-displacement specimens progressively occurred, it resulted in a ‘load plateau’ in the load-displacement response. response. Both the failure process and the final damage modes of the composite bolted joints deeply Both the failure process and the final damage modes of the composite bolted joints deeply affected the affected the captured Lamb wave feature in the Lamb wave inspection procedure, and discussions captured Lamb wave feature in the Lamb wave inspection procedure, and discussions on this aspect on this aspect will be carried out in the following work. will be carried out in the following work.

Materials 2017, 10, 652

10 of 17

Materials 2017, 10, 652 Materials 2017, 10, 652

10 of 17 10 of 17

L/W=1:1 L/W=2:1 L/W=1:1 L/W=3:1 L/W=2:1 L/W=3:1

14000 14000 12000 12000 10000 10000 8000 8000 6000 6000 4000 4000 2000 2000 0 00 0

L/W=1:1 L/W=3:2 L/W=1:1 L/W=2:1 L/W=3:2 L/W=2:1

Load/N Load/N

Load/N Load/N

7000 7000 6000 6000 5000 5000 4000 4000 3000 3000 2000 2000 1000 1000 0 00 0

5 5

10 15 20 10Deflection/mm 15 20

25 25

30 30

5 5

10

15

Deflection/mm 10 15 Deflection/mm

Deflection/mm (a) Case-A (b) Case-B (a) Case-A (b) Case-B Figure Figure7.7.Typical Typicalload-deflection load-deflectioncurves curvesof ofthe thecomposite compositebolted boltedjoints jointsin intensile tensile test. test. Figure 7. Typical load-deflection curves of the composite bolted joints in tensile test.

20 20

25 25

Figure 8. The final damage modes of the composite bolted joints after the tensile tests (In this figure, Figure 8. The final damage modes of the composite bolted joints after the tensile tests (In this figure, NFigure indicates net-tension, S indicates shear out, and B bolted meansjoints bearing failure mode). 8. The final damage modes of the composite after the tensile tests (In this figure, N N indicates net-tension, S indicates shear out, and B means bearing failure mode). indicates net-tension, S indicates shear out, and B means bearing failure mode).

Because the composite bolted joints after the tensile test were completely discontinuous and the Because the composite bolted joints after the tensile test were completely discontinuous and the LambBecause wave could not be captured PZTafter sensors the SHM we selected specimens over the composite bolted by joints the tensile testprocessing, were completely discontinuous and the Lamb wave could not be captured by PZT sensors the SHM processing, we selected specimens over 5Lamb kN and 10 kN tensile loads as research objects to establish quantitative connections between the wave could not be captured by PZT sensors the SHM processing, we selected specimens over 5 kN and 10 kN tensile loads as research objects to establish quantitative connections between the extracted the damage concludedconnections from the L-D curve the in 5extracted kN andsignal 10 kNcharacteristics tensile loads asand research objectsparameters. to establish As quantitative between signal characteristics and the damage parameters. As concluded from the L-D curve in Figure 7, these applied loads were very close to the failure tensile load of the joints. The corresponding extracted signal characteristics and theclose damage Asload concluded fromThe thecorresponding L-D curve in Figure 7, these applied loads were very to theparameters. failure tensile of the joints. damage modes of the specimens after these tensile loads are illustrated in Figure 9. Unlike the final Figure 7, these applied loads were very close to the failure tensile load of the joints. The corresponding damage modes of the specimens after these tensile loads are illustrated in Figure 9. Unlike the final damage modes in Figure 8, the damage modes are loads mainly failure mode damage specimens after these tensile arecomprised illustratedof inbearing Figure 9. Unlike the damage modes modes of in the Figure 8, the damage modes are mainly comprised of bearing failure mode accompanied with some shear out, and net-tension failure modes. The A0/S0 wave amplitude and final damage modes in Figure 8, the damage modes are mainly comprised of bearing failure mode accompanied with some shear out, and net-tension failure modes. The A0/S0 wave amplitude and their ratio tendency aftershear the tensile load are presented in modes. Figure 10. ItA0/S0 should be noted that the accompanied with some out, and net-tension failure wave amplitude andwave their their ratio tendency after the tensile load are presented in Figure The 10. It should be noted that the wave was excited and received by the PZT actuator-sensor pairs whose locations were the same as the ratio tendency after the tensile load are actuator-sensor presented in Figure It should be noted that wave was excited and received by the PZT pairs10. whose locations were thethe same aswas the specimens in Figure 5. In contrast to Figure 5, the amplitude in Figure 10 showed a decline in excited and in received actuator-sensor pairs locations were the as the specimens Figureby5.the In PZT contrast to Figure 5, thewhose amplitude in Figure 10same showed a specimens decline in tendency increase in L/W 5, ratio. in Figure with 5. In an contrast to Figure the amplitude in Figure 10 showed a decline in tendency with an tendency with an increase in L/W ratio. increase in L/W ratio.

Materials 2017, 10, 652

11 of 17

Materials 2017, 10, 652 Materials 2017, 10, 652

11 of 17 11 of 17

Figure 9. Damage modes of the composite bolted joints before completely failure after tensile test Figure 9. Damage Damage modes modes of of the the composite composite bolted bolted joints joints before before completely completely failure failure after after tensile tensile test test Figure (In this this figure, figure, N N indicates indicates net-tension, net-tension, SS indicates indicates shear shear out, out, and and BB means meansbearing bearingfailure failuremode). mode). (In (In 0.10 0.10

2.0 2.0

0.08 0.08

1.5 1.5

0.06 0.06

1.2 1.2

0.04 0.04

0.8 0.8

0.5 0.5

0.02 0.02

0.4 0.4

0.0 0.0

0.00 0.00

1.0 1.0

1.0 1.0

1.5 1.5

2.0 2.0Ratio L/W

L/W Ratio

2.5 2.5

3.0 3.0

S0 Mode S0 A0Mode Mode A0 Mode A0/S0 Ratio A0/S0 Ratio

2.0 2.0 1.6 1.6

Ratio Ratio

Ratio Ratio

2.5 2.5

Amplitude Amplitude

S0 Mode S0 A0Mode Mode A0 Mode A0/S0 Ratio A0/S0 Ratio

Amplitude Amplitude

0.18 0.18 0.16 0.16 0.14 0.14 0.12 0.12 0.10 0.10 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0.00 0.00

1.0 1.0

1.2 1.2

1.4 1.6 1.4 1.6 L/W Ratio

L/W Ratio

1.8 1.8

2.0 2.0

0.0 0.0

(a) Case-A (b) Case-B (a) Case-A (b) Case-B Figure 10. Waveform characteristics and the amplitude tendency in different composite bolted joints Figure 10. Waveform characteristics and the amplitude tendency in different composite bolted joints Figure 10. Waveform characteristics and the amplitude tendency in different composite bolted joints after tensile load. after after tensile tensile load. load.

4.3. Lamb Wave Inspection of the Composite Bolted Joints during Tensile Test 4.3. Test 4.3. Lamb Lamb Wave Wave Inspection Inspection of of the the Composite Composite Bolted Bolted Joints Joints during during Tensile Tensile Test Figure 11 shows the amplitude of S0 and A0 mode as the function of applied load in different Figure 11 shows the amplitude of S0 and A0 as of applied load in shows thetest. amplitude A0 mode mode as the the function function load in different different jointsFigure during11the tensile Clearly of in S0 theand figure, the amplitude showedofa applied declining tendency with joints during the tensile test. Clearly in the figure, the amplitude showed a declining tendency with joints duringof the tensile test.tensile Clearly in the figure, amplitude showed aindeclining with the increase the applied load. Taking thethe Lamb wave captured Case A astendency the example, the increase of the applied tensile load. Taking the Lamb wave captured in A the the of of theA0 applied tensile load. Taking theS0 Lamb in Case Case A as asapplied the example, example, the increase amplitude mode was larger than that of modewave whencaptured the tensile load was at the the amplitude of A0 mode was larger than that of S0 mode when the tensile load was applied at the amplitude of A0 mode was larger than that of S0 mode when the tensile load was applied at the the same level. As discussed above, this phenomenon is associated with the unique wave propagation same level. As discussed above, this phenomenon is associated with the unique wave propagation same level. As discussed this phenomenon is associated the unique wave propagation characteristics of A0 andabove, S0 mode in the composite plates. with Differences between Lamb wave characteristics ofA0A0 and S0 mode incomposite the composite plates. Differences between wave characteristics of and S0 mode in the plates. Differences between Lamb waveLamb amplitudes amplitudes of unloaded joints are very easy to distinguish between different L/W specimens, while amplitudes of unloaded joints are very easy to distinguish between different L/W specimens, while of joints arewas verynearly easy to between different L/W specimens, the S0/A0 theunloaded S0/A0 amplitude thedistinguish same when the joints were under tensile load.while In terms of the the S0/A0 amplitude was nearlywhen the same when the under joints were under load.the In joints termsinofCase the amplitude wasB,nearly the same the joints were load. tensile In terms joints in Case it seemed that the wave amplitude in L/Wtensile = 3:2 specimens wasof always larger than joints in Case B, it seemed that the wave amplitude in L/W = 3:2 specimens was always larger than B, seemed thatunder the wave amplitude in amplitude L/W = 3:2 specimens always larger than other cases theit other cases tensile load. The decreasingwas phenomenon could bethe interpreted as the other cases under tensile load. The amplitude decreasing phenomenon could be interpreted under tensile load. The decreasing phenomenon could be interpretedvary as the following: Asasit the following: As it is amplitude known, the Lamb wave propagation characteristics with entry angle, the following: As it wave is known, the Lamb wave propagation characteristics vary withand entry angle, is known, the characteristics vary angle, frequency frequency andLamb structural propagation geometry [35,36]. Compared thewith L-Dentry curve in Figure 7 and thestructural damage frequency and structural geometry [35,36]. Compared the L-D curve in Figure 7 and geometry [35,36].9,Compared themodes L-D curve inconsideration Figure 7 and the damage in Figure 9,the the damage damage modes in Figure the damage under were mainlymodes comprised of bearing failure, modes in Figure 9, the damage modes under consideration were mainly comprised of bearing failure, modes consideration weretomainly comprised of bearing failure, severity is proportional whose under severity is proportional the applied tensile load. With the whose increasing of tensile load, the whose severitytensile is proportional to the appliedoftensile load. the increasing of tensile load, the to the applied Withvarying the increasing tensile load, With the waveforms changed with varying waveforms changed load. with the geometries. Moreover, since tensile strength on thethe composite waveforms changed with the varying geometries. Moreover, since tensile strength on the composite geometries. sincetension, tensile strength on the composite bolted joints tension, bolted jointsMoreover, is offset axial the contact between the two plates in is theoffset jointaxial region beganthe to bolted joints is offset axial tension, the contact between the two plates in the joint region began to contact between the two plates in the joint region began to weaken. As a result, much less kinetic weaken. As a result, much less kinetic energy of from the wave particles could go through the joints weaken. As a result, much less kinetic energy of from the wave particles could go through the joints region and received by the PZT sensor. In order to verify this assumption, we performed the axial region and received by the PZT sensor. In order to verify this assumption, we performed the axial tensile experiment on a pristine composite plate and a composite plate with a bolt hole (drilled plate). tensile experiment on a pristine composite plate and a composite plate with a bolt hole (drilled plate).

Materials 2017, 10, 652

12 of 17

energy of from the wave particles could go through the joints region and received by the PZT sensor. 10, 652 12 of 17 In Materials order to2017, verify this assumption, we performed the axial tensile experiment on a pristine composite Materials 2017, 10, 652 12 of 17 plate and a composite plate with a bolt hole (drilled plate). Figure 12 shows the amplitude tendency Figure 12 shows the amplitude tendency of the two specimens as the function of the applied load. With of the two specimens as the function of the applied load. With the tensile load increasing from 0 to Figure 12 shows the amplitude tendency ofN, thethe two specimens as the function of thefrom applied load. With the tensile load increasing from 0 to 10,000 S0/A0 mode amplitudes declined 0.887/0.997 to 10,000 N, the S0/A0 mode amplitudes declined from 0.887/0.997 to 0.0243/0.0611 for the pristine plate the tensile loadfor increasing from 0 to 10,000 N, the amplitudes fromdeclined 0.887/0.997 to the pristine in Figure 12a,S0/A0 whilemode the S0/A0 modedeclined amplitudes from in 0.0243/0.0611 Figure 12a, while the S0/A0plate mode amplitudes declined from 0.74/0.843 to 0.024/0.0838 for the 0.0243/0.0611 for the pristine plate in Figure 12a, while the S0/A0 mode amplitudes declined from 0.74/0.843 to 0.024/0.0838 for the drilled plate in Figure 12b. The wave amplitudes show a decreased drilled plateto in0.024/0.0838 Figure 12b.for The wave amplitudes show12b. a decreased tendency with theaincreasing of 0.74/0.843 drilled platewhich in Figure wave amplitudes show decreased tendency with the increasing the of tensile load, was due The to the increased geometry length of the tensile load,with which due to of the increased geometry of the specimens. Another aspect, the tendency thewas increasing tensile load,attenuation which waslength due to increased geometry of the specimens. Another aspect, the amplitude level inthe Figure 11 was sharper length than that of amplitude attenuation level in Figure 11 was sharper than that of Figure 12. By comparison, this was specimens. Another aspect, this the amplitude attenuation in weakened Figure 11 was sharper than that of Figure 12. By comparison, was mainly attributedlevel to the contact property of the mainly attributed to the weakened contact property of thetocomposite plate contact in the joint region.of the Figure 12. By comparison, this was mainly attributed the weakened property composite plate in the joint region. composite plate in the joint region. 0.16

L/W=1:1 L/W=2:1 L/W=1:1 L/W=3:1 L/W=2:1 L/W=3:1

0.16 0.14

0.12 0.10

0.40 0.35 0.35 0.30

Amplitude Amplitude

Amplitude Amplitude

0.14 0.12

0.10 0.08 0.08 0.06 0.06 0.04 0.04 0.02

0.30 0.25 0.25 0.20 0.20 0.15 0.15 0.10 0.10 0.05

0.02 0.00

0.05 0.00

0.00

0

1000

0

1000

0.05

2000

3000

Load/N 2000 3000

(a) Case-ALoad/N S0 mode (a) Case-A S0 mode

4000

5000

4000

5000

0.00

L/W=1:1 L/W=3:2 L/W=1:1 L/W=2:1 L/W=3:2 L/W=2:1

0.05 0.04 0.04 0.03

Amplitude Amplitude

Amplitude Amplitude

L/W=1:1 L/W=2:1 L/W=1:1 L/W=3:1 L/W=2:1 L/W=3:1

0.40

0.03 0.02 0.02 0.01 0.01 0.00 0.00

0

2000

0

2000

4000

6000

Load/N 6000 Case-BLoad/N S0 mode 4000

8000

10000

8000

10000

0.14 0.14 0.12 0.12 0.10 0.10 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0.00 0.00

0

1000

2000

3000

4000

5000

0

1000

Load/N 2000 3000

4000

5000

0 0

(b) Case-ALoad/N A0 mode (b) Case-A A0 mode

2000 2000

4000 6000 4000Load/N 6000

L/W=1:1 L/W=3:2 L/W=1:1 L/W=2:1 L/W=3:2 L/W=2:1

8000 8000

10000 10000

1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0

S0 Mode A0 Mode S0 Mode A0 Mode

0.9 0.9 0.8

S0 Mode A0 Mode S0 Mode A0 Mode

0.8 0.7 0.7 0.6

Amplitude Amplitude

Amplitude Amplitude

(c) (d) Case-BLoad/N A0 mode (c) Case-B S0 mode (d) Case-B A0 mode Figure 11. Comparison between amplitude of composite bolted joints during the tensile tests. Figure 11. Comparison between amplitude of composite bolted joints during the tensile tests. Figure 11. Comparison between amplitude of composite bolted joints during the tensile tests.

0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0.0

0 0

2000 2000

4000 6000 Load/N 4000 6000

8000 8000

10000 10000

0.0

0 0

2000 2000

4000 6000 8000 4000Load/N 6000 8000 Plate with bolt hole Load/N

10000 10000

(a) Pristine plate (b) Load/N (a) Pristine plate (b) Plate with bolt hole Figure 12. Amplitude of S0 and A0 mode of a pristine and drilled composite plate during the axial Figure 12. Amplitude of S0 and A0 mode of a pristine and drilled composite plate during the axial tension Figure 12.test. Amplitude of S0 and A0 mode of a pristine and drilled composite plate during the axial tension test. tension test.

Materials 2017, 10, 652 Materials 2017, 10, 652

13 of 17 13 of 17

5. Simulation Works on Lamb Waves Waves in in Composite Composite Bolted Bolted Joints Joints 5.1. 3D FE Model Description As discussed in the experimental section, the Lamb wave propagation phenomenon in the joint structural discontinuity. discontinuity. In order to further interrogate the joint region is very complex, mainly due to structural influence on the Lamb wave propagation feature, FE simulations For the the sake of influence wave propagation simulations were conducted. For brevity, the main brevity, main aim of the performed FE simulation was to study the Lamb wave propagating feature 1:1composite compositebolted boltedjoints jointsin inCase CaseA. A. In In [37], [37], we we carried carried out out the simulation work on woven in L/W L/W ==1:1 composite laminates under low-velocity impact load. Also based on the developed three-dimensional FE method in previous work [38], we adopted 3D Hashin criteria and a user material subroutine to perform the Lamb wave inspection on the composite bolted joints. The simulation was carried out on Abaqus/Explicit (Version: (Version: Abaqus Abaqus 6.12, 6.12, Simulia, Simulia, Providence, RI, USA) platform, and the FE model the Abaqus/Explicit with details details is is shown shown in in Figure Figure13. 13.The Thebolted boltedcomposite compositelaminates laminateshad hadfour fourlayers, layers,each eachlayer layerhaving havinga ◦ /0◦ ] , and the final dimension, bolt location thickness of 0.55 mm. The fiber stacking sequence was [90 a thickness of 0.55 mm. The fiber stacking sequence was [90°/0°] 4, and the final dimension, bolt 4 in the model the same as the experiment in Figurein 1 and Table 1. The laminates were assigned location in thewas model was the same as the experiment Figure 1 and Table 1. The laminates were anisotropic propertiesproperties accordingaccording to the parameters listed in Tables 3–5. For the assigned anisotropic to the parameters listed in Tables 3–5.interface For the property interface in the overlapped jointed region, contact’ was applied betweenbetween the twothe parts. terms the property in the overlapped jointed‘hard region, ‘hard contact’ was applied twoIn parts. In of terms interlaminar property betweenbetween each composite layer, thelayer, surfaces tied together the interaction of the interlaminar property each composite thewere surfaces were tiedintogether in the module in Abaqus. The element type was antype eight-node C3D8R with a spatial resolution 0.1 mm. interaction module in Abaqus. The element was an eight-node C3D8R with a spatial of resolution This moreThis thanisten times smaller thansmaller the smallest wave lengthwave of thelength Lambofwaves in thewaves considered of 0.1ismm. more than ten times than the smallest the Lamb in the frequency range [35,39]. For the excitation signal, a 3.5 cycle Hanning-windowed sinusoidal tone considered frequency range [35,39]. For the excitation signal, a 3.5 cycle Hanning-windowed burst was excited by awas circular finebymesh regionfine with diameter 10 mm. In order tomm. keepIn pace with sinusoidal tone burst excited a circular mesh regionofwith diameter of 10 order to the experiment, twoexperiment, steps were performed in theperformed simulation: pre-tightening on the boltforce was keep pace with the two steps were in athe simulation: aforce pre-tightening applied in the step, and then the Lamb wave behavior in the bolted jointsinwas on the bolt wasfirst applied in the first step, and thenpropagating the Lamb wave propagating behavior thecarried bolted out in was the second joints carriedstep. out in the second step.

Figure 13. FE model in the Lamb wave inspection simulation with details (Coordinate system: Figure 13.2:FE model and in the Lamb direction). wave inspection simulation with details (Coordinate system: 1: 1: length, thickness 3: width length, 2: thickness and 3: width direction).

5.2. Simulation Results 5.2. Simulation Results Unlike Lamb waves in isotropic plates, the wave velocity was subject to the propagation Unlike waves inmaterials isotropic plates, wave velocity was to the propagation direction direction in Lamb non-isotropic [40,41].the Therefore, before FEsubject simulations on composite bolted in non-isotropic materials [40,41]. before simulations on composite joints were joints were carried out, we firstlyTherefore, investigated the FE slowness profiles of [90°/0°]4bolted and [+45°/−45°] 4. ◦ /0◦ ] and [+45◦ /−45◦ ] . Figure 14 carried out, we firstly investigated the slowness profiles of [90 4 in the composite 4 laminates Figure 14 described the measured velocities of the S0 and A0 modes described the800 measured thefound S0 andthat A0 the modes the composite laminates (dimension: (dimension: × 800 × 2velocities mm3). Weofcan two in Lamb wave modes travelled at distinct 3 ). We can found that the two Lamb wave modes travelled at distinct velocities in 800 × 800 × 2 mm velocities in different directions. Taking [90°/0°]4 composite plate as the example, the S0 and A0 wave different directions. [90◦ /0◦ ]4were composite plate as the example, the S0 and A0 velocities velocities along 45° Taking fiber direction the lowest among all the directions. Thiswave phenomenon ◦ along 45 the fiberwave direction were the lowest all the directions. phenomenon matched thecan wave matched propagating theory among in laminated compositeThis panel in Equation (5), and be propagating theory in laminated composite panel in Equation (5), and can be explained as following: explained as following: The woven fabrics in the WGF/epoxy composite laminates are along 90°/0°, and the strength and modulus along the fiber direction was much higher in the composites. The particle elastic motions of Lamb wave in the fiber direction were stronger and this led to a high wave

Materials 2017, 10, 652

14 of 17

The woven fabrics in the WGF/epoxy composite laminates are along 90◦ /0◦ , and the strength and modulus along the fiber direction was much higher in the composites. The particle elastic motions of Materials 2017, 10, 652 14 of 17 Lamb wave in the fiber direction were stronger and this led to a high wave amplitude and velocity in the slowness profiles. be noted that the Lamb wavesthat in the same amplitude and velocity in It theshould slowness profiles. It should be noted the joints Lamb followed waves in the joints ◦ in slowness profiles in Figure 14a. Since the sensor-actuator path was along the fiber direction (0 followed the same slowness profiles in Figure 14a. Since the sensor-actuator path was along the fiber ◦ slowness (0° profile), the captured Lamb wave signal hadwave a higher amplitude and velocity in the /0◦ ]4 direction in slowness profile), the captured Lamb signal had a higher amplitude and[90 velocity composite bolted joints. bolted joints. in the [90°/0°] 4 composite

(a) [90°/0°]4

(b) [+45°/−45°]4

Figure 14. Slowness profiles of S0 and A0 modes in WGF/epoxy composite laminates with different Figure 14. Slowness profiles of S0 and A0 modes in WGF/epoxy composite laminates with different ply modes. ply modes.

The wave structures of the composite bolted joints in the simulation and experiment are compared The wave structures of theiscomposite bolted joints the simulation experiment are compared in Figure 15. Because the wave mainly propagated byin particle motions and along the thickness direction, in Figure 15. Because the wave is mainly propagated by particle motions along the thickness we used the U2 displacement fields to represent the Lamb wave in the simulation. It wasdirection, obvious we used themode U2 displacement fields to represent thethe Lamb wave in and the simulation. It was obvious that that the S0 was predicted successfully, and simulation experimental results matched the S0Like mode predicted and the and experimental results matched well. well. thewas received signalsuccessfully, in experiments, twosimulation wave packages were found between S0 and A0 in Like the received signal in experiments, two wave packages were found between S0 and A0 in the simulation. Minor differences between the simulation and the experimental results of the the A0 simulation.were Minorattributed differences the experimental results of the A0 amplitude amplitude tobetween the fact the thatsimulation the elasticand vibration of A0 wave at the composite plate were attributed to the fact that the elastic vibration of A0 wave at the composite plate edgesdue in the edges in the experiment is difficult. Meanwhile, the distortion of the A0 wave packets to experimentcould is difficult. Meanwhile, distortionobserved. of the A0 As wave packets by duethe to relationship dispersion could also dispersion also have led to thethe differences calculated between have led to the differences As calculated by the between propagation propagation duration and observed. the sensor-actuator distance in relationship the FE, the predicted S0/A0 wave duration velocity and the sensor-actuator distance in the FE, the predicted S0/A0 wave velocity followed the dispersion followed the dispersion curves in Figure 4. Figure 16 shows the Lamb wave propagation feature curvesitin Figure 4. Figure 16the shows LambAs wave propagation feature whenS0 it and propagates through when propagates through jointthe region. can be seen, the fundamental A0 modes were the joint region. As can be seen, the fundamental S0 and A0 modes were captured well, and captured well, and their propagation duration was consistent with the signal curve in Figure 15.their It is propagation duration the signal curve in Figure 15. 16: It isExpansion worth mentioning that worth mentioning thatwas two consistent significantwith phenomena were found in Figure of the excited two significant phenomena found in Figure 16: Expansion of the excited Lamb waves occurred Lamb waves occurred at thewere incident edge of the jointed plate, and when it was propagating from a at the incident edge of the jointed plate, and when it was propagating from a large cross-section to a large cross-section to a small area at the second edge, reflection and transmission occurred. This small area at the second edge, reflection and transmission occurred. This phenomenon implies that the phenomenon implies that the wave motion in the composite bolted region was constructed during wave motion in the composite bolted region was constructed during the propagation. the propagation.

Materials 2017, 10, 652

15 of 17

Materials 2017, 10, 652 Materials 2017, 10, 652

15 of 17 15 of 17

Normalized Normalized Amplitude Amplitude

1.0 1.0

Experiment FE Simulation U2 Experiment FE Simulation U2 S0 mode 0.5 S0 mode 0.5 Incident wave Incident wave

A0 mode A0 mode

0.0 0.0 -0.5 -0.5 -1.0 -1.0 0 0

30 30

60

90

Time 60 (microsec) 90 Time (microsec)

120 120

150 150

Figure 15. Comparison of the normalized amplitude between FE simulation and the experiments. Figure 15. Comparison of the normalized amplitude between FE simulation and the experiments. Figure 15. Comparison of the normalized amplitude between FE simulation and the experiments.

Figure 16. Displacement fields (U2) of Lamb waves in the composites bolted joints for an excitation Figure 16. Displacement fields in (U2) waves in the composites bolted joints for an excitation frequency 210 kHz obtained theof FELamb simulations. Figure 16. of Displacement fields (U2) of Lamb waves in the composites bolted joints for an excitation frequency of 210 kHz obtained in the FE simulations. frequency of 210 kHz obtained in the FE simulations.

6. Conclusions 6. Conclusions 6. Conclusions This paper investigated Lamb wave-based structural health monitoring technology on two types This paper investigated Lamb wave-based structural health monitoring technology two types of composite bolted joints with different L/W ratios. As a preliminary study, the followingon conclusions This paper investigated Lamb wave-based structural health monitoring technology on two types of composite bolted joints with different L/W ratios. As a preliminary study, the following conclusions can be summarized: of composite bolted joints with different L/W ratios. As a preliminary study, the following conclusions can be summarized: summarized:between S0/A0 wave and the failure mode of the composite bolted joints were can be Relationships  Relationships between S0/A0mainly wave and the the failure mode of of theLamb composite were Damages in the joints amplitude wave,bolted and thejoints difference • found. Relationships between S0/A0 waveaffected and the failure mode ofthe the composite bolted joints were found. Damages in the joints mainly affected the amplitude of the Lamb wave, and the difference in S0/A0Damages amplitude could be adopted to identify the failure of mode of thewave, joints and afterthe tensile load. found. in the joints mainly affected the amplitude the Lamb difference in S0/A0 amplitude could bewave adopted to identify failuretorque mode of the tensile load.  The amplitude of the Lamb varied with thethe applied the joints joints.after The tensile S0/A0 wave in S0/A0 amplitude could be adopted to identify the failure mode on of the joints after load.  The amplitude of theaLamb wave varied with the applied torque on the joints. The S0/A0 showed decrease tendency the increasing of the tensile load, thiswave was • amplitudes The amplitude of the Lamb waveinvaried withwith the applied torque on joints. Theand S0/A0 wave amplitudes showed a decrease in tendency with the increasing of tensile load, and this was mainly caused by the increasedinspecimen and the weakened contact in the joint amplitudes showed a decrease tendencylength with the increasing of tensile load,property and this was mainly mainly caused by the increased specimen length and the weakened contact property in the joint region tensilespecimen test. causedduring by the the increased length and the weakened contact property in the joint region region during the tensile test.capture the slowness profiles of Lamb waves well when they were  The built 3D FE model could during the tensile test.  The built 3D in FEmulti-layered model could composite capture thelaminates. slowness The profiles of Lamb waves when they were propagating captured wave in thewell simulation matched • The built 3D FE model could capture the slowness profiles of Lamb waves well when they were propagating in multi-layered composite laminates. The captured wave in the simulation matched the experiment results, and wave expansion and reflection phenomenon were observed in the propagating in multi-layered composite laminates. The captured wave in the simulation matched the experiment results, and wave expansion and reflection phenomenon were observed in the overlapped joint region. the experiment results, and wave expansion and reflection phenomenon were observed in the overlapped joint region. overlapped joint region. Acknowledgments: This work was supported by China Postdoctoral Science Foundation (No. 2016M601522) Acknowledgments: work Funds was supported by China Postdoctoral Science Foundation (No. 2016M601522) and the FundamentalThis Research for the Central Universities (No. 222201714015). Acknowledgments: work was supported by China Postdoctoral Science Foundation (No. 2016M601522) and and the FundamentalThis Research Funds for the Central Universities (No. 222201714015). Author Contributions: BinFunds Yang, for Fu-Zhen Xuan Universities and Yanxun(No. Xiang conceived and designed the experiments; the Fundamental Research the Central 222201714015). AuthorZhu Contributions: Binperformed Yang, Fu-Zhen Xuan and Yanxun and designed the Xiaojun experiments; Wujun and Jichao Xu the experiments; Dan Li,Xiang Kang conceived Yang, Chengqiang Luo and Tang Wujun Zhu Jichao Xu performed the experiments; Dan Li, Kang Yang, Chengqiang andthe Xiaojun analyzed theand data; Fu-Zhen Xuan contributed reagents/materials/analysis tools; Bin YangLuo wrote paper.Tang analyzed the data; Fu-Zhen Xuan contributed reagents/materials/analysis tools; Bin Yang wrote the paper.

Materials 2017, 10, 652

16 of 17

Author Contributions: Bin Yang, Fu-Zhen Xuan and Yanxun Xiang conceived and designed the experiments; Wujun Zhu and Jichao Xu performed the experiments; Dan Li, Kang Yang, Chengqiang Luo and Xiaojun Tang analyzed the data; Fu-Zhen Xuan contributed reagents/materials/analysis tools; Bin Yang wrote the paper. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2. 3. 4.

5.

6. 7. 8. 9. 10. 11.

12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.

Lobel, T.; Holzhuter, D.; Sinapius, M.; Huhne, C. A hybrid bondline concept for bonded composite joints. Int. J. Adhes. Adhes. 2016, 68, 229–238. [CrossRef] Fu, M.; Mallick, P.K. Fatigue of hybrid (adhesive/bolted) joints in SRIM composites. Int. J. Adhes. Adhes. 2001, 21, 145–159. [CrossRef] Vallee, T.; Tannert, T.; Meena, R.; Hehl, S. Dimensioning method for bolted, adhesively bonded, and hybrid joints involving fiber reinforced polymers. Compos. Part B 2013, 46, 179–187. [CrossRef] Esmaeili, F.; Chakherlou, T.N.; Zehsaz, M. Investigation of bolt clamping force on the fatigue life of double lap simple bolted and hybrid (bolted/bonded) joints via experimental and numerical analysis. Eng. Fail. Anal. 2014, 45, 406–420. [CrossRef] Samaei, M.; Zehsaz, M.; Chakherlou, T.N. Experimental and numerical study of fatigue crack growth of aluminum alloy 2024-T3 single lap simple bolted and hybrid (adhesive/bolted) joints. Eng. Fail. Anal. 2016, 59, 253–268. [CrossRef] Vijaya Kumar, R.L.; Bhat, M.R.; Murthy, C.R.L. Evaluation of kissing bond in composite adhesive lap joints using digital image correlation: Preliminary studies. Int. J. Adhes. Adhes. 2013, 42, 60–68. [CrossRef] Ostachowicz, W.M. Damage detection of structures using spectral finite element method. Comput. Struct. 2008, 86, 454–462. [CrossRef] Su, Z.Q.; Ye, L. Identification of Damage Using Lamb Waves; Springer: Berlin, Germany, 2009. Harb, M.S.; Yuan, F.G. Non-contact ultrasonic technique for Lamb wave characterization in composite plates. Ultrasonics 2016, 64, 162–169. [CrossRef] [PubMed] Kang, T.; Lee, D.H.; Song, S.J.; Kim, H.J.; Jo, Y.D.; Cho, H.J. Enhancement of detecting defects in pipes with focusing techniques. NDT E Int. 2011, 44, 178–187. [CrossRef] Gianneo, A.; Carboni, M.; Giglio, M. Feasibility study of a multi-parameter probability of detection formulation for a Lamb waves-based structural health monitoring approach to light alloy aeronautical plates. Struct. Health Monit. 2017, 16, 225–249. [CrossRef] Nayfeh, A.H. Wave Propagation in Layered Anisotropic Media: With Applications to Composites; Elsevier: New York, NY, USA, 1995. Yu, L.; Bottai-Santoni, G.; Giurgiutiu, V. Shear lag solution for tuning ultrasonic piezoelectric wafer active sensors with applications to Lamb wave array imaging. Int. J. Eng. Sci. 2010, 48, 848–861. [CrossRef] Yu, X.D.; Ratassepp, M.; Rajagopal, P.; Fan, Z. Anisotropic effects on ultrasonic guided waves propagation in composite bends. Ultrasonics 2016, 72, 95–105. [CrossRef] [PubMed] Pant, S.; Laliberte, J.; Martinez, M.; Rocha, B. Derivation and experimental validation of Lamb wave equations for an n-layered anisotropic composite laminate. Compos. Struct. 2014, 111, 566–579. [CrossRef] Zhou, S.; Wang, Z.Q.; Zhou, J.S.; Wu, X.D. Experimental and numerical investigation on bolted composite joint made by vacuum assisted resin injection. Compos. Part B 2013, 45, 1620–1628. [CrossRef] Ascione, F.; Feo, L.; Maceri, F. An experimental investigation on the bearing failure load of glass fiber/epoxy laminates. Compos. Part B 2009, 40, 197–205. [CrossRef] Ascione, F.; Feo, L.; Maceri, F. On the pin-bearing failure load of GFRP bolted laminates: An experimental analysis on the influence of bolt diameter. Compos. Part B 2010, 41, 482–490. [CrossRef] Feo, L.; Marra, G.; Mosallam, A.S. Stress analysis of multi-bolted joints for FRP pultruded composite structures. Compos. Struct. 2012, 94, 3769–3780. [CrossRef] Ahmad, H.; Crocombe, A.D.; Smith, P.A. Strength prediction in CFRP woven laminate bolted double-lap joints under quasi-static loading using XFEM. Compos. Part A 2014, 56, 192–202. [CrossRef] Matt, H.; Bartoli, I.; Lanza di Scalea, F. Ultrasonic guided wave monitoring of composite wing skin-to-spar bonded joints in aerospace structures. J. Acoust. Soc. Am. 2005, 118, 2240–2253. [CrossRef] Sherafat, M.H.; Guitel, R.; Quaegebeur, N.; Lessard, L.; Hubert, P.; Masson, P. Guided wave scattering behavior in composite bonded assemblies. Compos. Struct. 2016, 136, 696–705. [CrossRef]

Materials 2017, 10, 652

23. 24. 25. 26. 27. 28.

29.

30. 31. 32. 33.

34.

35. 36.

37.

38.

39. 40. 41.

17 of 17

Ren, B.; Lissenden, C.L. Ultrasonic guided wave inspection of adhesive bonds between composite laminates. Int. J. Adhes. Adhes. 2013, 45, 59–68. [CrossRef] Samaratunga, D.; Jha, R.; Gopalakrishnan, S. Wave propagation analysis in adhesively bonded composite joints using the wavelet spectral finite element method. Compos. Struct. 2015, 122, 271–283. [CrossRef] Deng, Q.T.; Yang, Z.C. Propagation of guided waves in bonded composite structures with tapered adhesive layer. Appl. Math. Model. 2011, 35, 5369–5381. [CrossRef] Geetha, G.K.; Mahapatra, D.R.; Gopalakrishnan, S.; Hanagud, S. Laser Doppler imaging of delamination in a composite T-joint with remotely located ultrasonic actuators. Compos. Struct. 2016, 147, 197–210. [CrossRef] Rucka, M. Experimental and numerical study on damage detection in an L-joint using guided wave propagation. J. Sound Vib. 2010, 329, 1760–1779. [CrossRef] Yang, B.; Zhang, J.F.; Zhou, L.M.; Lu, M.K.; Liang, W.Y.; Wang, Z.Q. Effect of fiber surface modification on water absorption and hydrothermal aging behaviors of GF/pCBT composites. Compos. Part B 2015, 82, 84–91. [CrossRef] Yang, B.; Wang, Z.Q.; Zhou, L.M.; Zhang, J.F.; Tong, L.L.; Liang, W.Y. Study on the low-velocity impact response and CAI behavior of foam-filled sandwich panels with hybrid facesheet. Compos. Struct. 2015, 132, 1129–1140. [CrossRef] Rose, J.L. Ultrasonic Waves in Solid Media; Cambridge University Press: New York, NY, USA, 1999. Pavlakovic, B.; Lowe, M. Disperse User’s Manual, version 2.0; Imperial College London: London, UK, 2003. Staszewski, W.J.; Lee, B.C.; Mallet, L.; Scarpa, F. Structural health monitoring using scanning laser vibrometry: I. Lamb wave sensing. Smart Mater. Struct. 2004, 13, 251–260. [CrossRef] Giurgiutiu, V. Predictive methodologies for the design of lamb-wave piezoelectric wafer active sensors for structural health monitoring, damage detection and failure prevention. In Proceedings of the 2009 NSF Engineering Research and Innovation Conference, Honolulu, HI, USA, 22–25 June 2009. Yu, L.; Momeni, S.; Godinez, V.; Giurgiutiu, V.; Ziehl, P.; Yu, J. Dual mode sensing with low-profile piezoelectric thin wafer sensors for steel bridge crack detection and diagnosis. Adv. Civ. Eng. 2012. [CrossRef] Ng, C.T.; Veidt, M. A Lamb-wave-based technique for damage detection in composite laminates. Smart Mater. Struct. 2009, 18, 074006. [CrossRef] Veidt, M.; Ng, C.T. Influence of stacking sequence on scattering characteristics of the fundamental anti-symmetric Lamb wave at through holes in composite laminates. J. Acoust. Soc. Am. 2011, 129, 1280–1287. [CrossRef] [PubMed] Yang, B.; Wang, Z.Q.; Zhou, L.M.; Zhang, J.F.; Liang, W.Y. Experimental and numerical investigation of interply hybrid composites based on woven fabrics and PCBT resin subjected to low-velocity impact. Compos. Struct. 2015, 132, 464–476. [CrossRef] Yang, B.; Xuan, F.Z.; Chen, S.J.; Zhou, S.P.; Gao, Y.; Xiao, B. Damage localization and identification in WGF/epoxy composite laminates by using Lamb waves: Experiment and simulation. Compos. Struct. 2017, 165, 138–147. [CrossRef] Yelve, N.P.; Mitra, M.; Mujumdar, P.M. Detection of delamination in composite laminates using Lamb wave based nonlinear method. Compos. Struct. 2017, 159, 257–266. [CrossRef] Wandowski, T.; Malinowski, P.H.; Ostachowicz, W.M. Circular sensing networks for guided waves based structural health monitoring. Mech. Syst. Signal Process. 2016, 66–67, 248–267. [CrossRef] Su, Z.Q.; Ye, L.; Lu, Y. Guided Lamb waves for identification of damage in composite structures: A review. J. Sound Vib. 2006, 295, 753–780. [CrossRef] © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).