Experimental validation of a novel piezoelectric ...

Energy 153 (2018) 882e889

Contents lists available at ScienceDirect

Energy journal homepage: www.elsevier.com/locate/energy

Experimental validation of a novel piezoelectric energy harvesting system employing wake galloping phenomenon for a broad wind spectrum Muhammad Usman a, *, Asad Hanif b, **, In-Ho Kim c, Hyung-Jo Jung c a b c

School of Civil and Environmental Engineering, National University of Science and Technology, Sector H-12, Islamabad, Pakistan Department of Civil Engineering, Mirpur University of Science and Technology, Allama Iqbal Road, Mirpur, AJK, Pakistan Department of Civil and Environmental Engineering, Korea Advanced Institute of Science and Technology, Yuseong-gu, Daejeon, Republic of Korea

a r t i c l e i n f o

a b s t r a c t

Article history: Received 19 December 2017 Received in revised form 12 February 2018 Accepted 18 April 2018 Available online 18 April 2018

In this paper, a novel piezoelectric energy harvesting system using the wake galloping phenomenon is explored for the broad wind spectrum. Wake galloping is an aerodynamic instability phenomenon which has promising potential in energy harvesting. The offered advantage by the proposed system is having a wider wind speed range for a reliable and sustainable energy source for a small wireless sensor node, in addition to being simple and easily applicable to civil structures. In the proposed system, the flow of wind runs parallel to the placed cylinders with upstream cylinder fixed at one end while the downstream one is placed over an unimorph cantilever beam with piezoelectric film attached to it. To validate the effectiveness of the proposed system, several tests were conducted against wind speeds approaching till 10 m/s with varying cylinder spacing. The results revealed an optimum spacing between two cylinders of 3D and the cut-in speed was estimated to be 4 m/s. The attained results were smartly analyzed to further explore the effect of cylinder spacing on the aerodynamic vibrations and the other associated parameters. © 2018 Elsevier Ltd. All rights reserved.

Keywords: Piezoelectric sensor Energy harvester Wake galloping Wind tunnel test

1. Introduction Aerodynamic instability has posed serious challenges to civil engineers with the evolution of large-scale structures, especially high-rise buildings and long-span bridges. The uncertain and variable wind properties in the real environmental conditions and the congested urban development pattern make it one of the complex problems to solve. Large and unwanted wind vibrations caused by aerodynamic instability have raised major concerns and have been thoroughly investigated over last many decades [1e6]. These phenomena include vortex induced vibration, flutter, buffeting and galloping. In practical applications, more than one of these phenomena may occur at the same time depending on the geometry of the structure and the wind flow properties. The issue of aerodynamic instability first came to limelight in 1940 when the old

* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (M. Usman), [email protected] (A. Hanif). https://doi.org/10.1016/j.energy.2018.04.109 0360-5442/© 2018 Elsevier Ltd. All rights reserved.

Tacoma Narrows Bridge in the United States of America collapsed due to torsional flutter [1]. The fundamentals of this phenomenon were later studied by Duncan [7] followed by numerous other researchers investigating more or less the similar scenarios [8e10]. In contrast to the structural instability issue, the aerodynamic instability can also be seen as an opportunity to harness useful energy based on wind power. The divergent vibration properties of this phenomenon can be used to power small wireless sensor nodes in addition to adversely affecting the serviceability of the large structures. Wireless sensor nodes have a major power concern, which has been the focus of researchers for several years [11e14]. Several power-generating devices were proposed based on flutter and other aerodynamic instability phenomena [15e20]. Among the energy harvesting devices based on such aerodynamic phenomena, the most important parameter is the limited viable range of wind speeds. Based on this parameter, various aerodynamic phenomena were compared and the wake galloping turned out to be most suitable for an energy harvesting system, because of its low cut-in speed and wider wind speed range [21]. Wake galloping is basically defined as the divergent vibrations of the downstream cylinder caused by the wake from the upstream

M. Usman et al. / Energy 153 (2018) 882e889

cylinder. When a single cylinder with flexible base is placed in the wind and it vibrates, the vibration of the cylinder is called the galloping phenomenon. While when another cylinder with a fixed base is placed on the upstream side of the flexible base cylinder, its vibrations increase significantly due to the re-attachment of wakes from the upstream cylinder onto the downstream cylinder. This increased vibration of the downstream cylinder is termed as wake galloping [10]. Jung and Lee [21] proposed a new electromagnetic energy harvesting system based on the wake galloping phenomenon. They carried out characterization test on the wake galloping phenomenon and also proposed a system containing a permanent magnet and solenoid coil system in order to generate electrical power. Their system was tested against maximum wind speeds of up to 8e17 m/ s for different configurations of cylinders. For the spacing of 3D they tested up to 10 m/s, however for the larger spacing they tested for wind speed of up to 16 m/s. They recommended a cut-in speed of 4.5 m/s. Although they tested their system for a wide range of wind speeds and validated the effectiveness of their system, still their system contains several limitations. Although their proposed system provided an efficient way of harnessing the wind-based vibration energy, they could not develop a practicable configuration of their system, which could be easily installed on a civil engineering structure. Moreover, the presence of an electromagnetic system on the structure might cause an additional vibration issue. Recently, Abdelkefi et al. [22] proposed another galloping and wake galloping based energy harvesting system. They mainly focused on a single piezoelectric beam with a square section tip mass subjected to the galloping phenomenon. They also mentioned a similar system subjected to wake galloping aerodynamic instability, in which they recommended to consider different parameters such as upstream cylinder diameter, the spacing between two cylinders, the flow speed, and the load resistance. They, however, did not try to design an optimum system based on wake galloping. The main reason why they did not proceed with the wake galloping energy harvesting system is that they only considered the low wind speed ranges. For low-speed range, the energy harvester based on galloping phenomenon performed better than the wake galloping based system. Additionally, they found that the upstream cylinder did not vibrate at all at low wind speeds. The important factor that seems missing is associated with the wind speeds that are unpredictable and vary in a very wide range. Therefore, the system designed for low wind speed might not be stable and even get damaged in higher speed ranges. For the piezoelectric energy harvesting systems based on simple galloping phenomenon, there is a major limitation of wind speed range to be less than 2.5 m/s [20]. Jung and Lee [21] have also developed awake galloping based energy harvesting system. However, this system also possesses certain reservations that need to be addressed. In their system, two parallel cylinders were used. The upstream cylinder was termed as a dummy and the downstream cylinder was taken as the main test cylinder. The downstream cylinder was connected to springs, permanent magnet and the solenoid coil. In this arrangement, the vibrations of the downstream cylinder caused the permanent magnets to oscillate with reference to the solenoid coil which in turn caused the generation of the useful energy, which can be used, for instance, to power a set of small wireless sensor nodes mounted on a cable-stayed bridge. Although apparently the system looked simple and effective, its implication on large scale still needed serious clarifications. The proposed system was difficult to be mounted on a real structure such as a cable-stayed bridge, which ultimately reduces the effectiveness of the electromagnetic device. Furthermore, the wind load of the additional structure would impart an additional load on the

883

original structure and might be the reason to induce additional vibrations. 1.1. Research significance In the current research, an optimum energy harvesting system based on wake galloping using an unimorph piezoelectric beam has been proposed. The proposed system effectively addresses the aforementioned literature gaps in much effective and sustainable way by encompassing a wider range of wind speeds. The proposed system offers many advantages over the existing wake galloping based energy harvesting systems. It is easy to mount over a structure and needs minimum maintenance because of minimum vibration compared with the previously proposed system Jung and Lee [21]. Also, the adjustments of the cylinders can be made flexible in the proposed system such that the downstream cylinder is always placed on the backside of the upstream cylinder resulting in the elimination of any possible additional wind loads. The proposed system is much suitable for medium and high wind speed ranges in contrast to the previously developed system by Abdelkefi et al. [20] limited to low wind speeds. The optimum distance between the two cylinders has also been effectively evaluated in this study based on the set of experiments. In literature, the usual practice is the use of square cross-section that seem much suitable to the simple galloping based system [23], whereas in the parallel cylinder arrangement the square cross section cause more stability issue rather than ensuring sustainable vibrations. The proposed system has a configuration which is easier to implement and the size of the system is small which eliminates the possibility of causing additional vibrations to the main structure. 2. Proposed system An effective piezoaeroelastic energy harvesting system based on wake galloping phenomenon has been successively proposed in this work. For both the upstream and the downstream cylinders, circular cross-sections of similar diameters were considered. As mentioned in the literature that the upstream cylinder does not vibrate enough in the presence of a downstream cylinder, therefore, the upstream cylinder was designed as rigid with the energyharvesting piezoelectric system only applied at the downstream cylinder in the system proposed. The downstream cylinder was

Fig. 1. Schematic diagram of the proposed system.

884


mounted on top of an unimorph piezoelectric cantilever beam as a tip mass. The schematic diagram of the proposed system is shown in Fig. 1. In the wake galloping aerodynamic phenomenon, the wake is created from the upstream cylinder and when this wake tries to reattach to the downstream cylinder, it causes the vibration of the downstream cylinder. The force exerted on the downstream cylinder can be represented by two orthogonal components i.e., drag and lift components, respectively. The drag component is the one which has the same direction as the direction of wind flow, while the lift force is exerted in the direction perpendicular to the wind flow direction. The vibration of the downstream cylinder being studied is mainly caused by the lift component of the force. The downstream cylinder is placed on a thin cantilever beam, upon which a piezoelectric layer is pasted. The piezoelectric layer when undergoes strain due to vibration of cantilever beam generates voltage which can be harvested by connecting it to an electric circuit. The proposed system is studied by analyzing the results of a campaign of experimental tests. The wind tunnel tests were carried out for a total of 60 cases. The wind speeds were taken from 1 m/s to 10 m/s and the spacings of cylinders were varied from 1.73D to 6D, where D is the diameter of the cylinder. Different wind speeds were considered to find out the cut-in speed and the wind speed range where the proposed system shows efficient performance. In addition, different values of spacing were considered in order to find out the optimum spacing between the two cylinders, where the proposed system shows efficient performance. To verify the practicability and stability of the system several wind tunnel tests were performed. Two parallel cylinders were placed in the ways to facilitate the wind flows parallel to them as shown in Fig. 1. In the literature, different cross-sectional geometries of the cylinders have already been analyzed but the current study focuses the use of circular geometry as the wake galloping phenomenon is mainly known to occur for the civil structures with cylindrical cross section. Non-circular cross-sections do have higher aerodynamic vibrations at low wind speeds, but they are not very suitable for an energy harvesting system with wide applicable wind speed range [23]. The upstream cylinder was made of a light plastic material having a mass per unit length of 1 kg/m. The upstream cylinder was rigidly attached to the floor of the wind tunnel apparatus so that the wake galloping vibrations of the downstream cylinder can be maximized. The downstream cylinder was half in length compared to that of the upstream cylinder, rest of the length consisted of an unimorph cantilever beam made of stainless steel with a macro fiber composite (MFC) layer attached to it. The downstream cylinder was fixed as half so as to provide the maximum surface area for the lift force to act, meanwhile providing enough length to the unimorph beam to vibrate significantly. The downstream cylinder was hollow from inside and had lower mass density compared to the upstream cylinder. The fixation of all the parts was done using tightened bolts. The P2 type MFC product (M2814-P2) was purchased from a company called Smart Material Corporation. The MFC is basically an innovative actuator, which contains aligned rectangular fibers of the piezoceramic material. These fibers in the product we used were aligned along the width of the cantilever beam. These films usually come as a package which is sealed and ready to use, which can be easily attached to any surface or embedded in a layer. The unimorph beam under the downstream cylinder was such oriented that it can vibrate in the direction perpendicular to the direction of wind flow as elaborated in Fig. 1. The diameter of both the cylinders was kept same as suggested by Zdravkovich [10] and the diameter was taken as 3 cm to keep it similar to the system studied by Jung and Lee [21]. The length of the cylinders was taken as 250 mm in order to keep the blockage ratio at 8%. The blockage

ratio is the ratio of obstructed area to the total cross sectional area of the wind tunnel, and it should be less than 10% in order to provide enough air flow for studying any aerodynamic behavior [24]. Other geometric details of the proposed system are summarized in Table 1. The placement of downstream cylinder was also kept varying relative to the upstream one to determine its optimum position for different levels of wind flow. The analyzed positions of the downstream cylinders are briefly summarized in Fig. 2. Where L designates the center-to-center distance between the two cylinders along the direction of the wind. The second cylinder is basically mounted over a thin and flat cantilever beam. Hence, half the length of the downstream cylinder is replaced by the cantilever beam. In this study, we considered the spacing between the two cylinders in the range of 1.73D to 6D, where D is the diameter of both the cylinders. In the experimental setup, the minimum possible spacing was 5.3 cm which comes out to be 1.73D. The wind speeds were considered up to 10 m/s. 3. Experimental set-up Experimental tests were carried out at Structural control and Intelligent Systems (SCAIS) lab in KAIST using the wind tunnel test equipment shown in Fig. 3(a). A uniform cross-section wind tunnel was used with fairly uniform wind speed. The wind tunnel used for the experiment is designed to eliminate the effect of boundary layer effect at the entrance, by providing the several hexagonal openings separated by a honeycomb shaped very thin layers of aluminum foil. Also, the smooth wall surface of wind tunnel and the placement of test sample is such that it minimizes the boundary layer thickness. The experimental setup for this study included a wind speed sensor, a displacement sensor, and a device to measure the voltage across the MFC film. The voltage responses across the MFC film were measured using an acquisition device (PXI-4461: National Instruments). The speed sensor was used to ensure the constant wind speed in each case. The wind speed was controlled with a knob provided at the wind tunnel apparatus (Fig. 3(b)), while the actual wind speed was measured by the wind speed sensor shown in Fig. 3(c). The wind speed sensor was placed at the center of the air flow channel and the proposed system was also placed along the centerline of the channel so that the measured wind speed is the same as the one applied to the model of the proposed system. The displacement sensor was used to measure the displacement of the cantilever beam in the direction perpendicular to the wind flow. The voltage was measured against time for each case. In order to eliminate the noise from the sensor signals, a

Table 1 Characteristics of the proposed system. Parameter

Unit

Value

Length of upstream cylinder Diameter of both cylinders Mass of upstream cylinder Tip mass of downstream cylinder Length of the cantilever beam Width of the cantilever beam Thickness of the cantilever beam MFC Film Characteristics Piezoelectric coefficient (d31) Poisson's ratio (v12) Operational bandwidth Young's Modulus Shear Modulus Epoxy Epo-Tek 353ND Properties Operating Temp. Range Viscosity (mixed)

mm mm g g mm mm mm

250 30 249 45 85 30 0.3

pC/N kHz GPa GPa

210 0.31 0e10 15.857 5.515

Degree Celcius Cps

50~þ200 3000e5000


885

cases. The voltage and displacement were measured for all the cases. The downstream cylinder was fixed over a simple cantilever beam using bolts and the cantilever was also tightly fixed to the base with bolted connections. Heat-curing epoxy Epo-Tek 353ND manufactured by Epoxy Technology Inc [25]. was used to attach the MFC film to the beam surface. Epo-Tek 353ND is a two-component epoxy which is ideal for bonding metals, ceramics and plastics. Some of its properties are provided in Table 1. Additionally, a cable was soldered at both ends of the MFC film in order to measure the voltage across the film during the vibration. The voltage was measured across the MFC film for all the cases. The measurements were made for wind speeds of up to 10 m/s. 4. Experimental results Fig. 2. Different positions of downstream cylinder considered relative to the upstream cylinder.

The number of cases considered is shown in Table 2. The time history data were collected for each case for a given uniform wind speed. As the wind speed increases, the stability of the vibration also increased considerably. For each time-history result, the peak factor is computed and interpreted along with the apparent shape of the plot and the amplitude values. The peak factor can be derived in terms of statistical parameters of moving mean and RMS response [16]. In simpler words, it can be defined as the ratio of maximum fluctuating component of mean to its standard deviation. However, for ease of computations, it has been found to be approximately the same as the ratio of the peak and RMS values of any time signal [26]. In this study, a simplified approach to compute the peak factor for each of the time histories was used as given in Eq (1) below:

Peak Factor ¼

Fig. 3. (a) Wind tunnel equipment. (b) Airspeed controller. (c) Air-speed sensor.

low-pass filter was applied. The details of the proposed system made for the experiment are given in Fig. 4. As evident in the figure, the upstream cylinder is connected with a rigid connection to the wind tunnel test equipment and the downstream cylinder is placed over a flexible cantilever beam with piezoelectric film mounted on it. A series of experiments were carried out for the mentioned

xmax xrms

(1)

where x can be the time history response of the displacement of the downstream cylinder we are considering. Here xmax is the maximum displacement amplitude and xrms is the root mean square value for the complete displacement vector. For an ideal sinusoidal time history, the peak factor is known to be exactly √2 and for buffeting response to the nonlinear transient winds, it is reported to be around 2.5 [16]. Fig. 5 shows the time histories of displacement for the case L ¼ 1.73D and wind speeds of 3 m/s, 7 m/s and 10 m/s. One time history each is represented from a low, medium and high range of wind speed. It is obvious that the vibration amplitude is increasing with the increasing wind speed. From the time histories above, it is evident that the vibration at the higher wind speeds is more stable and harmonic. The peak factor for the 3 m/s wind speed case is 2.93, while for the cases of 7 m/s and 10 m/ s the peak factor is 1.86 and 1.81, respectively. The cut-in speed is the critically important factor for the wake galloping based energy harvesting system, which can be defined by a speed beyond which we can have a significant increase in the vibration amplitudes. For two cylinders in tandem arrangement, the aerodynamic behavior is basically dependent on the spacing between the two cylinders. As reported by Ref. [10], the wake galloping system can be further subdivided into three different aerodynamic mechanisms. This subdivision basically depends on the location of the downstream cylinder with respect to the

Table 2 Investigated Experimental setups.

Fig. 4. Experimental setup: (a) Top view. (b) Front view.

Parameter

Setups

Number of setups

Total Cases: 60 Wind speed Spacing between both cylinders

1 m/se10 m/s 1.73 De6 D

10 6

886


upstream cylinder. For the spacing between two cylinders less than 2D, the free shear layers actually do not reattach to the downstream cylinder due to the very small spacing between both the cylinders; instead, they practically behave like a slender body. Therefore, the vortex formed at the downstream of both the cylinders is similar to the vortex formed around a slender rectangular body. This phenomenon is quite similar to the gap-flow-switch mechanism [27], although with the configurations studied in the work there might be very rare chances for this phenomenon to occur. If we look at the time histories in Fig. 5, the transitional flow behavior is obvious at the low wind speed which is also depicted by its peak factor. However, the flow of the medium and higher wind speeds were shown to be relatively smooth and stable which is also shown by a lower peak factor value. Fig. 6 shows the time histories for a 3D spacing and wind speeds of 3 m/s, 7 m/s and 10 m/s. Here also, it is clear that the vibration is increasing directly with the increasing wind speed. And similarly, there is some high-frequency response at the lower wind speeds. The peak factor for the 3 m/s case is 2.93, similar to the 1.73D case. The high-frequency response and the higher value of the peak

factor can be explained by the fact that for lower displacement amplitudes, the noise values in the measured signal can be more significant considering the very low values of the displacement signal itself. The peak factors for the 7 m/s and 10 m/s cases are 1.62 and 1.69 respectively. Here, the medium and high wind speed time histories are more stable and close to the standard harmonic vibration. As reported in the literature [10], the spacing between 2D and 3.8 D show the pure wake galloping phenomenon and the case of 3D spacing, shown in Fig. 6, should fall in this range. Although the peak factor for the maximum wind speed is slightly higher, this is explainable by the slight fluctuation in the middle of the time history. As it can be seen in Fig. 5(c) that between 4 and 5 s the amplitude suddenly dropped, which caused the peak factor to go slightly higher. If the spacing of the cylinders is further increased beyond 4D, the wake galloping phenomenon starts to fade and a transition towards the wake displacement method can be seen. Fig. 7 shows the time histories for the spacing of 5D and the wind speeds of 3 m/ s, 7 m/s and 10 m/s. Here the trend of increasing displacement amplitude with increasing wind speed is similar to the trend in Figs. 5 and 6. The high-frequency time history plot for the wind speed of 3 m/s and the peak factor of 3.12 is higher compared to the same wind speed case in Figs. 5 and 6. However, in this case for 7 m/ s wind speed, the time history is not very smooth and the peak factor, in this case, is also higher at 2.44, compared to the same wind speed cases in previous figures. This can be explained by the existence of the wake displacement phenomenon. For the 10 m/s case also the peak factor is relatively higher than the previous cases. Also, it can be noticed here that the maximum vibration amplitude is also considerably less than the 3D case. To investigate further, the power spectral density for the 3D spacing case was estimated. Fig. 8 shows the spectral density plot for wind speeds of 3 m/s, 7 m/s and 10 m/s. Fig. 8 acts as a guiding tool in the elaboration of spectrum summarized in Fig. 6, where for the wind speed of 3 m/s the high frequency was observed with higher peak factor of 2.93. If we correlate this peak factor with the 3 m/s curve in Fig. 8, we can see higher spectral densities for the frequencies of around 25 Hz in addition to the normal lower frequency where other curves also show higher spectral density. Additionally, the curves for the 7 and 10 m/s cases show a similar power spectral density curves with the peaks occurring at almost the same frequencies. Here all curves shows a peak at around

Fig. 6. The displacement time histories measured for the spacing L ¼ 3D (a) v ¼ 3 m/s. (b) v ¼ 7 m/s, and (c) v ¼ 10 m/s.

Fig. 7. The displacement time histories measured for the spacing L ¼ 5D (a) v ¼ 3 m/s. (b) v ¼ 7 m/s, and (c) v ¼ 10 m/s.

Fig. 5. The displacement time histories measured for the spacing L ¼ 1.73D (a) v ¼ 3 m/ s. (b) v ¼ 7 m/s, and (c) v ¼ 10 m/s.


100

3 m/s 7 m/s 10 m/s

90 80

Power Spectral Density

887

70 60 50 40 30 20 10 0

5

10

15

20 25 30 Frequency (Hz)

35

40

45

50

Fig. 8. Power spectral density for L ¼ 3D. Fig. 10. Maximum voltage generated for each case against wind speed.

frequency of 25 Hz, which shows presence of high frequency data, which may be because of electrical noise of the displacement sensors. Also the peak at frequencies less than 5 signify that the actual vibration response is almost of similar magnitude as that of noise. It endorses the conclusion made earlier that the highfrequency results are more prominent at the low wind speed. The maximum vibration amplitudes measured for each case are compiled in Fig. 9 against wind speed. This helps us understand the overall behavior of the system with different wind speeds. The voltage across the MFC film was also monitored to determine the extent of voltage generation. The maximum values of these voltages are plotted in Fig. 10. The trend of the increase with increasing wind speed is similar to the one noticed in Fig. 9. In Figs. 9 and 10 a clear transition from galloping to wake galloping and then from wake galloping to the wake displacement is evident. Starting from the case of 1.73D where the high cut-in speed is around 8 m/s and high peak values represent clear transitional phase from the simple galloping phenomenon to the wake galloping vibrations. High maximum values represent the effect of the wake galloping and high cut-in speed shows the effect of the galloping phenomenon. Moving forward to 2D and 3D cases, it can be clearly seen that the vibrations are transiting towards the wake galloping vibrations that are characterized by lower cut-in speeds with high and stable peak

Fig. 9. Maximum vibration amplitude for each case against wind speed.

values for a wider range of wind speeds. For the spacing range of 2D to 3.8D, the shear layers from the upstream cylinder are directly reattached onto the downstream cylinder. This phenomenon causes the wake galloping vibrations to occur and behind the downstream cylinder, only single vortex street is formed. For wake galloping phenomenon in cylinders of the same diameter, it is known that the critical spacing is approximately 3.8D [10], although this critical spacing may vary from system to system. The presented results also follow the same trend that the cut-in wind speed is consistently reducing. The cut-in speed for the case of spacing on 2D is 6 m/s, and that for 3D is around 4 m/s. Although this value is the same for the 4D case, the peak values for the 4D case have now started reducing. Hence in the results, the critical spacing is clearly visible at the 3D level. For the spacing values of 4D onwards we can see that the peaks are consistently reducing and also there is a wind speed range in each case with considerably constant vibrations. In this area, the free shear layers from the upstream cylinder actually fail to reach the downstream cylinder. Instead, a vortex is formed towards the upstream side of the downstream cylinder, which in turn causes two vortices behind the downstream cylinder. Such vortices are called the binary vortex [10] because they contain the effect from both the upstream and downstream cylinders. This phenomenon is called the wake displacement as it does not show any clear cut-in speed as is the case for the wake galloping phenomenon. In order to further investigate the effect of the cylinder spacing, a plot of maximum generated voltage against the spacing between the cylinders is presented in Fig. 11. Regarding the effect of wind speed as evident from Figs. 9 and 10 as well, the generated voltage is increasing with the proportionate increase in the wind speed. This increase becomes more prominent beyond the wind speed of 4 m/s. Moreover, the spacing between the two cylinders has also a significant effect on the wake galloping vibration. The voltage generation is maximum at the spacing of L ¼ 3D, with a consistent decrease in the values from 4D onwards. An important aspect to note in Fig. 11 is that the effect of critical spacing is only prominent in the wind speed range of 4 m/s to 7 m/s. For the wind speed values below 4 m/s, the voltage is very less and also it is almost independent of the spacing. Same is the case for the wind speeds beyond 7 m/s with an exception of 1.73D case at 7 m/s where the vibration was significantly lesser than other values reported for the same wind speed. For the higher wind speeds, there is a consistent decrease in the generated voltage with a constant increase of spacing between the two cylinders. This fact can be explained by

888


6. Concluding remarks

Fig. 11. Maximum voltage generated for each case against spacing.

the consistent weakening of the binary vortices mentioned earlier with increasing distance in the downstream direction. 5. Discussion and further analyses The experimental results show visible transition among the different types of flow behaviors. The wake galloping vibrations are shown in medium wind speeds, and the critical spacing for the proposed system came out to be 3D. For low wind speeds, minimal vibrations were observed; on the other hand, for the high wind speed values, the high vibrations were noticed for most of the cases. These results seem quite logical, as for wind speed values of less than 4 m/s, the vibrations were not very significant because in these cases the wakes of the air are not strong enough to cause a vibration of the downstream cylinder and vibrations were quite similar to the case of a single slender body. This might explain the fact that [22] did not expand the scope of his work on the proposed system. They studied the wake galloping based energy harvesting system for wind speeds of 3.9 m/s and compared the results with the galloping based energy harvester. As mentioned before, for the wind speeds less than 4 m/s, the proposed system also behaved similar to a single slender body. Therefore, they did not see the prospect in wake galloping based energy harvesting system because they were studying the system in a wind speed range which is not suitable for wake galloping vibrations. Most of the results show the similar trend compared to the work reported by Lee [27]. Considering that the proposed system addresses well the limitations of the existing system, therefore, it appears to be more practical in nature. In this study, it was determined that the spacing of 3D between the two cylinders is the optimum spacing for the wake galloping based energy harvesting system. Also, the cut-in speed was determined to be around 4 m/s, beyond which the wake galloping vibrations occur, therefore, the proposed system is an energy harvesting device that can work in medium to large wind speeds. Jung and Lee [21] have already done a detailed discussion on the possible application of wake galloping based energy harvesting system in real civil engineering structures. Wireless sensor nodes are widely being used for structural health monitoring application, especially for the cable stayed bridge structures [28,29]. Proposed system has a great potential to address the energy issue being faced by the most of structural health monitoring devices [21].

A wake-galloping based vibration-based energy harvester is proposed using the MFC piezoelectric film. This is a very simple yet reliable method to harness the naturally available wind energy. The proposed system exhibits a significant improvement compared to the previously proposed wake galloping energy harvesting system. The proposed system is easy to install and has no stability issue in higher wind speeds. The proposed system has obvious advantages over the existing systems, such as it has a very wide range of wind speeds and also the system is more suitable for practical applications. In the current study we have verified that if we study wake galloping based system in broader wind speed range it in fact gave better energy harvesting performance. The usefulness of the proposed system was verified by carrying out the lab scale tests for various cases. The effect of wind speed and the relative location of the downstream cylinder on the generated voltage has been determined in this study. The results show that the proposed system performs well for the wind speeds higher than 4 m/s and the optimum location of the downstream cylinder came out to be equal to three times the diameter of the upstream cylinder. Different types of aerodynamic vibration behaviors were observed for different spacing between the cylinders. When two cylinders were very near, the vibration pattern showed the transition from simple galloping to wake galloping. For medium spacing values, the pure wake galloping phenomenon was observed, while for higher spacing values, the wake displacement phenomenon was observed. The wake galloping energy harvesting system has a lot of potential applications, especially in the large-scale civil engineering structure, whose integrity (i.e., structural health) has to be monitored across its lifespan. The issue of powering wireless sensor nodes, on such structures which are located at locations difficult to access, can be resolved by using this sustainable energy harvesting system. Acknowledgments This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (NRF2013R1A1A2011351, NRF-2015R1A2A1A16074926) and a grant from the Korean Minister of the Ministry of Land, Infrastructure and Transport (MOLIT) as 'U-City Master and Doctor Course Grant Program'. References [1] Diego C, Angel CA. Improving the wind stability of suspension bridges during construction. J Struct Eng 2001;127(8). [2] Agar TJA. The analysis of aerodynamic flutter of suspension bridges. Comput Struct 1988;30(3):593e600. [3] Yongxin Y, Yaojun G, Zhao L. Aerodynamic investigation on flutter instability of the first sea-crossing bridge in ChinaeThe east sea bridge. In: Proceeding of the 4th European and African conference on wind engineering, Czech; 2005. [4] Yongxin Y, Yaojun G, Haifan X. Aerodynamic flutter control for typical girder sections of long-span cable-supported bridges. J Wind Struct 2009;12(3). 2009. [5] Yongxin Y, Yaojun G. Equivalent static wind loading for long-span bridges. In: Proceeding of the 3rd symposium on new strategy for wind disaster risk reduction of wind sensitive infrastructures, China; 2010. [6] Yongxin Y, Yaojun G. Equivalent static wind loading for long-span arch bridges. In: Proceeding of the 4th symposium on new strategy for wind disaster risk reduction of wind sensitive infrastructures, Japan; 2011. [7] Duncan WJ. The fundamentals of flutter. Aircraft Eng Aero Technol 1945;17(2):32e8. [8] Davenport AG. Buffeting of a suspension bridge by storm winds. J Struct Div 1962;88(3):233e70. [9] Thomas DG, Kraus KA. Interaction of vortex streets. J Appl Phys 1964;35(12): 3458e9. [10] Zdravkovich M. Review of interference-induced oscillations in flow past two parallel circular cylinders in various arrangements. J Wind Eng Ind Aerod

M. Usman et al. / Energy 153 (2018) 882e889 1988;28(1e3):183e99. [11] Roundy S, Wright PK, Rabaey J. A study of low level vibrations as a power source for wireless sensor nodes. Comput Commun 2003;26(11):1131e44. [12] Gungor VC, Hancke GP. Industrial wireless sensor networks: challenges, design principles, and technical approaches. IEEE Trans Ind Electron 2009;56(10):4258e65. [13] Sim S-H, et al. A wireless smart sensor network for automated monitoring of cable tension. Smart Mater Struct 2013;23(2), 025006. [14] Kuo Y-W, Li C-L. Design of long range low power sensor node for the last mile of IoT. In: Consumer Electronics-Taiwan (ICCE-TW), 2016 IEEE international conference on. IEEE; 2016. [15] Bryant M, Garcia E. Development of an aeroelastic vibration power harvester. In: Proc. SPIE; 2009. [16] Chen X. Prediction of buffeting response of long span bridges to transient nonstationary winds. In: Proceedings of the seventh international colloquium on bluff body aerodynamics & applications (BBAA7). Shanghai, China; 2012. [17] Sousa V, et al. Enhanced aeroelastic energy harvesting by exploiting combined nonlinearities: theory and experiment. Smart Mater Struct 2011;20(9), 094007. [18] Ali M, Arafa M, Elaraby M. Harvesting energy from galloping oscillations. In: Proceedings of the world congress on engineering; 2013. [19] Erturk A, et al. On the energy harvesting potential of piezoaeroelastic systems.

889

Appl Phys Lett 2010;96(18), 184103. [20] Abdelkefi A, et al. Incident flow effects on the performance of piezoelectric energy harvesters from galloping vibrations. Theor Appl Mech Lett 2014;4(2). [21] Jung H-J, Lee S-W. The experimental validation of a new energy harvesting system based on the wake galloping phenomenon. Smart Mater Struct 2011;20(5), 055022. [22] Abdelkefi A, et al. Performance enhancement of piezoelectric energy harvesters from wake galloping. Appl Phys Lett 2013;103(3), 033903. [23] Kim H. Mechanism of wake galloping of two circular cylinders. Nagoya, Japan: Nagoya University; 2009. [24] Choi CK, Kwon DK. Wind tunnel Blockage effects on aerodynamic behavior of bluff body. Wind Struct 1998;1(4):351e64. [25] http://www.smart-material.com/MFC-product-main.html, [M.F]. [26] Chen X. Extreme value distribution and peak factor of crosswind response of flexible structures with nonlinear aeroelastic effect. J Struct Eng 2014;140(12), 04014091. [27] Lee SW. Energy Harvesting system based on wake galloping phenomenon. Dissertation. Korea Advanced Institute of Science and Technology; 2013. [28] Cho SJ, et al. Decentralized system identification using stochastic subspace identification for wireless sensor networks. Sensors 2015;15(4):8131e45. [29] Spencer BF, et al. Next generation wireless smart sensors toward sustainable civil infrastructure. Prec Eng 2017;171:5e13.