PaperThe following article is Open access

A multi-stable rotational energy harvester for arbitrary bi-directional horizontal excitation at ultra-low frequencies for self-powered sensing

, , and

Published 8 August 2024 © 2024 The Author(s). Published by IOP Publishing Ltd
, , Citation Sayed N Masabi et al 2024 Smart Mater. Struct. 33 095017 DOI 10.1088/1361-665X/ad649b

0964-1726/33/9/095017

Abstract

A rotational multi-stable energy harvester has been presented in this paper for harnessing broadband ultra-low frequency vibrations. The novel design adopts a toroidal-shaped housing to contain a rolling sphere magnet which absorbs mechanical energy from bidirectional base excitations and performs continuous rotational movement to transfer the energy using electromagnetic transduction. Eight alternating tethering magnets are placed underneath its rolling path to induce multi-stable nonlinearity in the system, to capture low-frequency broadband vibrations. Electromagnetic transduction mechanism has been employed by mounting eight series connected coils aligned with the stable regions in the rolling path of the sphere magnet, aiming to achieve greater power generation due to optimized rate of change of magnetic flux. A theoretical model has been established to explore the multi-stable dynamics under varying low-frequency excitation up to 5 Hz and 3 g acceleration amplitudes. An experimental prototype has been fabricated and tested under low frequency excitation conditions. The harvester is capable of operating in intra-well, cross-well, and continuous rotation mode depending on the input excitation, and the validated physical device can generate a peak power of 5.78 mW with 1.4 Hz and 0.8 g sinusoidal base excitation when connected to a 405 Ω external load. The physical prototype is also employed as a part of a self-powered sensing node and it can power a temperature sensor to get readings every 13 s on average from human motion, successfully demonstrating its effectiveness in practical wireless sensing applications.

Export citation and abstract BibTeX RIS

Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 license. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

1. Introduction

Wireless sensor networks (WSN) are one of the crucial technologies used in structural and human health monitoring in the era of the Internet of Things (IoT) [1, 2]. Their influence in structural health monitoring can be seen in wind farms as a leading example, where their application enables real-time wireless monitoring and help detect faults in turbine tower units ahead of time, contributing to reducing maintenance costs and improving their lifespans [3]. WSN applications in human health monitoring can positively influence areas such as healthcare, emergency responses and entertainment using implantable or wearable body sensor network systems [4]. One of the major challenges in deploying remote sensors forming a WSN concerns their power supply [5, 6]. Batteries are primarily used as a way of powering wireless sensors, but they are commonly known to have limited lifespan [7]. Wired power supply for these sensors is also not a favourable solution since it would be impractical for moving bodies and remote, inaccessible locations. Hence, in recent years, significant focus has been given in introducing power autonomy in remote sensors through energy harvesting methods [8, 9]. Energy harvesters are devices designed for capturing ambient energy from sources such as solar, thermal and vibration and converting them into electricity using various transduction mechanisms. Kinetic energy harvesting is popular for health monitoring applications since vibration is available from most moving sensing surfaces and it can be captured easily from the source.

To harness these vibrations, the most commonly implemented transduction mechanisms include electromagnetic [10, 11], electrostatic [12, 13], triboelectric [14, 15] and piezoelectric [16, 17]. Electromagnetic transducers are often preferred for vibration energy harvesting, since they benefit from having lower impedance, relatively higher power output [18], easy fabrication and integration with sensing systems [19]. One of the key challenges of harnessing kinetic energy is that low and irregular vibrations are available from sources such as wind turbines and human motion. Structural vibrations from turbine towers are within a very low frequency range below 2.4 Hz [20, 21], while human motion frequencies can be as low as 5 Hz with amplitudes below 3 g from normal gait [22]. Significant contributions have been made in recent years in harnessing low-frequency structural vibrations using resonant oscillators. Li et al [23] explored the effects of magnetic arrays with different magnet shapes (cubic and triangular) and configurations (such as Halbach and alternating polarity) to improve the electromechanical coupling for electromagnetic energy harvesting. The research showed that the cubic alternating array variant performed optimally by generating 35.5 mW when subjected to 24 Hz and 1 g harmonic excitation matching its resonant frequency. Linear resonance employing an alternating magnetic array has also been reported in [24] where the effect of abrupt magnetic flux density change is investigated to show a relationship between the number of abrupt changes and power density. The optimal case is shown to harvest 284 mW output power under 20 Hz and 1 g excitation. Despite their advantages in resonant conditions, linear vibration energy harvesters tend to have a narrow operational bandwidth, which limits their ability to harness the broadband low frequencies present in the highlighted applications [25].

Nonlinear dynamics is often employed to enable energy harvesting across a broad bandwidth. Cubic or third-order polynomial spring-based nonlinearities are present in Duffing-type oscillators which can harness vibrations from a slightly broader bandwidth compared to linear resonance, and this technique has been studied in [2628]. Wang et al [29] presents a piezoelectric energy harvester showcasing duffing type hardening nonlinearity using a bow-shaped buckled beam configuration to harvest rotational energy for structural health monitoring in jet engines. Their design is shown to operate over a 22.5 Hz operational bandwidth, and maximum power of 78.87 mW is extracted from 35 Hz excitation. Despite their advantages, for ultra-low frequency broadband excitations, Duffing-type oscillators generally provide a marginal advantage over linear energy harvesters [30].

Mechanical frequency conversion is popular technique in energy harvesting, which is used to absorb wide bandwidth of frequencies by involving a secondary resonating system. Halim et al [31] proposed a frequency up-converted electromagnetic energy harvester that uses two spring-mass damper systems with magnetic masses to harness human motion and generates up to 0.96 mW average power. Li et al [32] proposed a complex design of a stacked piezoelectric resonator with two degrees of freedom with frequency up-conversion to extract 0.11 mW average power and was tested up to 26 Hz. Other low frequency examples include magnetic plucking-based piezoelectric harvesting from knee joint [33], stacked electromagnetic approach for swing motions [34], planetary geared eccentric pendulum-based mechanism for harnessing wrist motion [35] and broadband rotational applications [36]. The knee joint application presented by Kuang et al [33] targets excitation from gait with frequencies near 0.9 Hz and 57 angular displacement to harvest 5.8 mW output power. However, to achieve this, a ring formed of 8 piezoelectric bimorph beams is coupled to the shank and an outer ring with 32 plucking magnets is coupled to the thigh for energy extraction. The planetary geared mechanism design from Cai et al [35] is adaptable for ankles and wrists, and their prototype can harvest up to 2.95 mW from ankle swing motion at 1.2 Hz walking frequency.

A more effective approach involves introducing potential energy wells in the path of inertial resonators to create stable operating regions so that they can resonate within the potential wells or transfer between them when adequate energy is attained from the source vibrations. Depending on the number of potential wells, energy harvesters can be classified as mono-stable, bistable or multi-stable for harnessing broadband vibrations. Fan et al [37] presented a mono-stable electromagnetic energy harvester using magnetic end stoppers that delivers 1.15 mW peak power at 9 Hz. Piezoelectric mono-stable harvesting has been explored to harness vortex-induced vibrations [38], by employing magnetic repulsive mechanism [39] and using dual-cantilever structures [40, 41]. Bistable energy harvesting has been implemented to improve energy harvesting performance using plucking mechanism in piezoelectric beams with tip masses [4244], sliding impacts for triboelectric transduction [45] and using buckled beam strategy for electromagnetic transduction [46]. Fang et al [42] achieved bistability using springs in their piezoelectric-electromagnetic hybrid design to harness 2.98 mW from 7.5 Hz excitations with marginally broader bandwidth compared to linear resonance. Triboelectric transduction is used by Tan et al [45] where sliding-impact based bistable nonlinearity is employed and 148 µW is extracted from 6 Hz harmonic vibrations. Buckled beam piezoelectric harvester with bistability is used by Cottone et al [46] to achieve broadband response between 20–106 Hz, and this design achieves 8 mW maximum power from 48 Hz and 0.5 g amplitude excitations. Multi-stable energy harvesters (MEHs) have attracted significant attention in recent years due to their featured dynamics compared to the other methods available. By increasing the number of equilibrium positions, inter-well vibrations across multiple potential wells can lead to significantly higher output compared to mono-stable and bistable designs [47]. Quad-stable piezoelectric harvesting as an example is shown using springs by Yan et al [48] to harness wideband low frequency between 6–12 Hz to generate 1.4 mW maximum output during a forward frequency sweep up to 35 Hz with 0.6 g amplitude. Other notable advances in multi-stable systems for low frequencies include spring-based mechanisms [49] and designs that use magnetic forces [50, 51].

Adequate power generation still remains a challenge for real-time monitoring applications since vibration from low frequency sources such as turbine tower vibrations and human motion are random, multi-directional and broadband. Current solutions are limited by the complexity of the design and low-power extraction due to ineffective conversion of the available vibration from these sources. Hence, to address these challenges, this paper presents a novel electromagnetic nonlinear energy harvester that introduces multi-stability in a toroidal travelling path of a sphere magnet inertial mass, converting random bi-directional low frequency excitation into rotational motion. The looped path enables continuous rotational motion of the sphere magnet, due to the absence of physical barriers or end stoppers to restrict its motion. The multi-stable dynamics combined with looped rolling motion for bidirectional vibration harnessing is introduced and studied for the first time to harness ultra-low frequency broadband vibration, along with random human motion below 5 Hz. A theoretical model is developed and validated using experimental results. A parametric study is conducted to explore the effects of different excitation conditions and analyse the unique characteristics. The fabricated prototype is employed as a part of a self-powered sensing system to demonstrate its ability to provide power autonomy in WSN applications.

The remainder of the paper is organised as follows: section 2 contains the working principle and theoretical modelling of the MEH; section 3 shows the electromechanical behaviour of the designed energy harvesting system; section 4 details the prototype fabrication and experimental procedure and dynamic testing of the physical prototype; a self-powered sensing setup is powered using the proposed device in section 5; and finally, the conclusions and suggestions for future work are presented in section 6.

2. Energy harvester design and modelling

2.1. Concept and operating principle

The schematic of the proposed MEH is presented in figures 1(a)–(c). A sphere magnet is placed in a torus-shaped non-magnetic casing, which acts as a mechanism to convert linear vibration input or limb swing motion into rotational motion. The circumference of the rolling path in the torus casing is n times larger than that of the sphere magnet, where n is a positive integer. This allows the sphere magnet to complete n number of full revolutions during one complete cycle along the rolling path. A set of 2n tethering magnets are evenly distributed along this path as shown on figure 1(b). These magnets are of alternating polarity and at a distance of $\pi r_{m}$ relative to each other, where rm is the radius of the rolling magnet. Parallel to the tethering magnet array, same number of series-connected coils are placed above the rolling magnet's path, ensuring that the coils are centrally aligned with the tethering magnets.

Figure 1.

Figure 1. Schematic of the multi-stable energy harvester (a), demonstration of motion of the rolling magnet as it travels between potential wells created by the alternating tethering magnets (b), exploded view of the proposed harvester (c), self-powered applications in wind turbines and human motion (d), and IoT application in wind farms using the multi-stable energy harvester for self-powered structural health monitoring (e).

Standard image High-resolution image

The tethering magnets creates attractive forces in the rolling path of the sphere magnet, enabling multi-stability in the system. When the proposed design is exposed to input vibrations, the sphere magnet moves with rotational motion along the path, and current is induced through the coils by electromagnetic induction. Having the tethering magnets aligned with the coils ensures that the sphere magnet can start oscillating from near its vicinity when excitation is induced from its resting position. Additionally, having the tethering magnets placed at a distance of half the circumference of the sphere magnet with alternating polarity enables the sphere magnet to complete exactly a half rotation as it transitions between the potential wells created by the tethering magnets, as demonstrated in figure 1(b). This means that the sphere magnet can maintain rolling motion without slipping, and change its magnetization direction as it travels from coil to coil, in effect maximizing the rate of change of magnetic flux to enhance the energy harvesting performance. The potential applications for this proposed device, such as wind turbine tower condition monitoring or human health monitoring are given in figure 1(d). Self-powered sensor nodes using this proposed energy harvester can be used in an IoT WSN setup, as demonstrated in figure 1(e), where the energy harnessed can be used to power a sensor and a wireless transmission unit, and useful information such as acceleration and temperature can be conveyed remotely for structural health monitoring. With these applications in mind, the size of the device is limited to 100 mm span and a sphere magnet of 10 mm radius is employed as the inertial generator. As a result, n value is derived as 4.

In summary, the operation of the sphere magnet from one stable position to the next occurs as follows: The motion of the sphere magnet originates from one of the potential wells as the initial position when it is subjected to external forcing. At this stage, the magnetic flux is the highest in the vicinity of the coil placed above this potential well, since one of its poles are facing the coil where magnetic flux density is the highest. The motion of the sphere magnet is dependent on the relationship between the kinetic energy gained from the input vibrations and the potential energy from the tethering magnet causing it to remain in the well. If the energy absorbed is greater than that of the attractive potential well, it can conquer the well and continue travelling towards the following well with opposite polarity. As the sphere starts moving, vibration is converted into rotational motion and magnetic flux change occurs in the air gap. As it completes a half rotation, it reaches the following potential well created by the attractive forces from the next tethering magnet with reversed polarity. The opposite polarity between wells ensures the rolling without slipping motion of the sphere magnet, reducing losses due to friction and maximising the rate of change of magnetic flux. As the sphere magnet reaches the next potential well, its polarity is reversed as a result and this phenomenon continues along the looped rolling path.

Low frequency energy harvesting using rolling magnet mechanism has been explored in the past using looped path and ferromagnetic rail [52] or cylindrical magnets with diametrical magnetization [53], but low power output due to weak electromechanical coupling or energy losses due to mechanical limiters and damping constraints are challenges that need to be addressed. The toroidal path introduced in this design allows the inertial mass to achieve continuous unrestricted motion since the path is looped and without any physical barriers and the electromechanical coupling is enhanced by introducing top-mounted coils and alternating tethering magnets in the rolling path. The ratio of circumference of the rolling magnet and that of the effective rolling path in the casing allows the magnet to complete 4 full rotations during each cycle, allowing it to effectively utilise the area available.

2.2. Theoretical model

By considering the rolling motion of the sphere magnet on the circular path, the equation of motion can be derived using Newton's second law of motion as follows:

Equation (1)

where the mass and radius of the rolling magnet are m and rm respectively, c is the mechanical damping coefficient, θ is the angular displacement of the magnet, $F^{i}_\textrm{mag}$ is the sum of attractive forces from the tethering magnets ($i = 1,2,3, {\ldots} 8$), $\hat{\Theta}$ is the electromagnetic coupling factor, I is the current through the coils, ω is the excitation frequency, y is the excitation amplitude and the rotational inertia of the magnet is expressed as $J = \frac{2}{5}mr_{m}^{2}$. The MEH is considered as a voltage source and is connected to a load resistance $R_\textrm{l}$, therefore, the following equation is derived using Kirchhoff's voltage law:

Equation (2)

where L and $R_\textrm{c}$ are the coil's total internal inductance and resistance respectively. Equations (1) and (2) are coupled using $\hat{\Theta}$, which represents the rate of change of magnetic flux through the coils with respect to the sphere magnet's displacement. To derive this, the vertical component of magnetic flux density Bz present in the vicinity of the coil's midplane z is first modelled using cylindrical polar coordinates as follows:

Equation (3)

where µ0 is the permeability of free space, s is the distance the sphere magnet has rolled, φ and R are the angular position and radius of the rolling path. From equation (3), the coupling factor can be calculated as a derivative of the magnetic flux Φ through a coil with N turns as follows:

Equation (4)

where a1 and a2 are the inner and outer radius of the coil. Finally, the attractive magnetic forces can be expressed by considering the dipole moment interaction between the sphere magnet the tethering magnets:

Equation (5)

where ri is the distance between the centre of the $i_\textrm{th}$ tethering magnet and the sphere magnet and $m_\textrm{r}$ and $m_\textrm{t}$ are the dipole moments of the rolling and tethering magnets designed based on [54]. The distance ri can be expressed as:

Equation (6)

where $h_\textrm{rt}$ is the vertical gap between the rolling and tethering magnets.

3. Numerical analysis

3.1. Potential energy distribution

The electromechanical properties and dynamics of the MEH are studied by solving equations (1)–(6) numerically using MATLAB. The combined potential energy profile due to the addition of tethering magnets in the rolling path is illustrated in figure 2, showing the stable equilibrium positions and the attractive and repulsive regions. From figure 2(b), 8 stable and 8 unstable equilibria can be seen where the rolling magnet faces attractive and repulsive forces. The strength of the potential barriers can be varied by changing the vertical gap between the rolling and tethering magnets. While the unstable equilibrium positions ensure that the rolling magnet does not initiate its motion from these regions, the potential wells in the negative regions offer 8 stable positions in the rolling path and as a result the operational modes of the rolling magnet can change depending on the excitation conditions. The operation modes can vary based on the relationship between the kinetic energy of the sphere magnet and the potential energy from the tethering magnets.

Figure 2.

Figure 2. Potential energy distribution in the energy harvester rolling magnet's travelling path in polar (a) and cartesian (b) coordinates.

Standard image High-resolution image

3.2. Magnetic flux density distribution

To investigate the energy dissipated in the coils as a result of the sphere magnet's motion, the magnetic field is calculated. Figure 3 shows the combined magnetic field distribution in the vicinity of the coil region due to rolling motion of the sphere magnet in the provided path. The 8 tethering magnets are placed apart by a distance of half the perimeter of the rolling magnet, as shown by the peaks are visible at $\pi r_{m}$ displacement in the potential energy profile. By placing the coils centrally aligned to the point where peak amplitudes are shown, maximum voltage can be induced in the coils for harvester's enhanced generation capabilities. At zero displacement position, the vertical magnetic field is negative and maximum, illustrated by the blue tip in figure 3(b). As the sphere magnet travels a distance of $\pi r_{m}$ and is positioned at $\pi /4$ radians in the circular path, the polarity reverses and a positive maximum peak with the same amplitude as that of the previous position is achieved as shown in the yellow tip. Thus, rate of change of magnetic flux is optimised during operation as the sphere magnet passes each coil.

Figure 3.

Figure 3. Magnetic field distribution along the rolling path due to rolling magnet's motion in the energy harvester using polar plot (a) and effect of changing gap between rolling magnet and top-mounted coils (b).

Standard image High-resolution image

3.3. Dynamics of the system

The dynamic characteristics of the MEH under harmonic excitation is investigated to establish the operation modes as the input is varied. Figure 4 presents the voltage and displacement profiles with 2 Hz excitation frequency, generated using the numerical model developed in section 2.2. The results presented in this section have been solved numerically using MATLAB ode45 ordinary differential equation solver, based on explicit Runge–Kutta formulae of order 4 and 5. The step size is determined by the solver, but is generally fixed to 1 µs for the analysis. The duration for each simulation is set up to 1000 s to ensure steady state response is obtained.

Figure 4.

Figure 4. Multi-stable energy harvester dynamics assessed and three modes of operation determined using 2 Hz harmonic excitations. Displacement history, phase portrait and voltage for intra-well motion at 0.1 g (a)–(c), cross-well motion at 0.4 g (d)–(e), and continuous rotational motion at 0.7 g (g)–(i) are determined.

Standard image High-resolution image

When the MEH is excited with low energy, the kinetic energy gained by the inertial mass is weaker than the potential energy barrier, causing it to oscillate near its initial resting position. This type of motion is labelled as intra-well motion. Exposing the device to higher energy excitation input causes the sphere magnet to conquer one or more potential wells and travel with oscillatory motion, allowing more energy to be extracted through the transducers mounted above. This behaviour is labelled as cross-well motion. Increasing the excitation further causes the sphere to overcome all the potential wells as it accelerates until reaching a periodic regime, and it is characterised as continuous rotation. With a low amplitude excitation of 0.1 g, the dynamics in figures 4(a)–(c) shows that the sphere oscillates in intra-well motion. At this stage, the output voltage is quite low and does not exceed 22 mV, since the inertial mass oscillates near one stable equilibrium position. Cross-well motion is documented in figures 4(d)–(f) when 0.4 g excitation is given to the system, where output voltage of up to 1.2 V peak can be achieved across a fixed load of 400 Ω. In this mode, the system travels through potential wells and continues oscillatory motion, and higher energy is released compared to intra-well motion as the hopping phenomenon occurs. However, the energy provided is not high enough to sustain continuous rotational motion. As the energy increases slightly with a higher amplitude of 0.7 g in figures 4(g)–(i), continuous rotation is achieved and periodic AC voltage with 1.4 V peak is induced. Continuous rotational motion is the desired mode of operation for the proposed design, since energy from external excitation is absorbed effectively and the output voltage is adequate for power conditioning and storage.

4. Prototype and experimental study

4.1. Experimental Setup

An experimental prototype of the proposed energy harvester was fabricated to study the performance of the designed system. In figures 5(a)–(c), the three parts of the prototype are shown. The tethering magnets were attached to a disk, ensuring that they were alternating polarity relative to each other. The toroidal casing was fabricated in parts and then assembled to fit the sphere magnet inside as shown in figure 5(b) and the copper coils were placed on another circular disk. The parts were then assembled as shown in figure 5(d) ensuring that the tethering magnets and the coil centres were aligned. The casings were printed using 3D printing and the magnets were chosen as commercially available N42 grade NdFeB varieties. The parameters associated with the energy harvester prototype are presented in table 1. The completed prototype was attached to a slider controlled by a variable frequency drive, capable of generating base excitation up to 2.5 Hz and 1 g excitation amplitudes. The electric characteristic of the device was recorded using an oscilloscope as shown in figure 5(d) and the motion of the exposed device was recorded using a camera setup as presented in figure 5(e).

Figure 5.

Figure 5. Dissembled prototype: tethering magnet array (a), spherical rolling magnet fitted into toroidal casing (b), and plate containing series connected copper coils (c). Assembled device mounted on vibrating shaker for recording output voltage (d) and partially exposed prototype used for measuring displacement using high-speed camera (e).

Standard image High-resolution image

Table 1. Design parameters and material properties of the multi-stable rotational energy harvester.

SymbolDescriptionValue
m Rolling magnet mass29.4 g
ζ Mechanical damping coefficient0.0291 kg s−1
rm Rolling magnet radius10 mm
$r_\textrm{t}$ Tethering magnets' radii1.5 mm
$l_\textrm{t}$ Tethering magnet height2 mm
$h_\textrm{rt}$ Vertical gap between sphere and tethering magnet21 mm
$h_\textrm{rc}$ Vertical gap between sphere magnet and coil14 mm
a1 Coil inner width4 mm
a2 Coil outer width14 mm
a3 Coil outer length18 mm
R Radius of rolling path39 mm
$B_\textrm{r}$ Residual flux density1.3 T
L Total Coil inductance82.1 mH
$R_\textrm{c}$ Total Coil resistance405.7 Ω

4.2. Electrical performance

Figure 6(a) depicts the open circuit AC voltage from the prototype when exposed to varying range of base excitation amplitudes at 1.4 Hz. From the profile, it is apparent that the output voltage jumps from 1.1 V to 2 V peak value when the amplitude is increased from 0.2 g to 0.3 g. The relationship will be explored further in the following section. To determine the maximum power achievable, resistive loads between 100 Ω–800 Ω were placed across the harvester while excited at 1.4 Hz and 0.8 g vibrations. As highlighted in figure 6(b), a maximum peak power of 5.78 mW was extracted when external impedance matching the internal resistance of the coils was connected to the device.

Figure 6.

Figure 6. Open circuit output voltage at 1.4 Hz with 0.1 g –1 g amplitudes (a) and output Voltage and Power graphs for varying external loads (b) to find maximum power output.

Standard image High-resolution image

The results in figure 7 are obtained with a 405 Ω load connected across the harvester, matching the internal resistance. In figure 7(a), a clear transition between intra-well to full revolution can be seen as excitation amplitude raised from 0.3 g to 0.4 g at 1.4 Hz, but at higher frequencies the voltage response is more chaotic. In figure 7(b), at 0.3 g the sphere magnet starts to operate in the cross-well mode and at 0.4 g the motion changes between the three operation modes defined in section 3.1. The effect on output voltage due to different operation modes can be analysed using figures 8(a)–(c), where the output voltage is derived due to the harvester being excited with 1.9 Hz and 0.1 g, 0.4 g and 0.8 g acceleration amplitudes respectively. Figure 8(a) shows steady state intra-well response due to 1.9 Hz 0.1 g input excitation as 0.18 V low voltage peaks are generated from the harvester. The energy received by the sphere magnet is not enough to conquer the potential energy well it originates from, causing it to oscillate near one stable equilibrium. Increasing the amplitude to 0.4 g reveals a more chaotic response, as the output voltage profile changes abruptly without reaching steady state. The regions marked A, B and C can be used to describe the characteristics of this mode of operation. From A to C, the voltage profile shows a transition from a few cycles of low amplitude peaks to an envelope of high amplitude output in B before transitioning back to lower amplitude profile in C. This behaviour continues during its operation with a 0.4 g amplitude. This transition indicates that the energy absorbed by the sphere magnet is enough to occasionally escape an initial potential well and hop across a few corresponding wells, but not sufficient to sustain continuous rotational motion. Finally, at 0.8 g, the energy gained by the sphere magnet is enough to continue conquering the potential wells present due to the tethering magnets and sustain steady state rotational motion, as apparent in figure 8(c).

Figure 7.

Figure 7. Output voltages at 1.4 Hz (a) and 1.9 Hz (b) at different excitation amplitudes.

Standard image High-resolution image
Figure 8.

Figure 8. Harvester operation characterized into intra-well (a), chaotic cross-well (b) and full revolution (c) modes, obtained from 1.9 Hz harmonic excitation with 0.2 g, 0.4 g and 0.8 g acceleration amplitudes, respectively.

Standard image High-resolution image

4.3. Harvester motion profile

Using the camera setup shown in figure 5(e), the motion of the sphere magnet during operation was recorded as shown in figure 9. To gain visual access to the motion, the coil plate was removed to partially expose the roof of the MEH to record the effect of potential barriers (videos showing three operating modes in figure 9 are available in the supplementary files). Figures 9(a)–(c) can provide insight for the output behaviour in figures 8(a)–(c) respectively, which was extracted in cartesian coordinates from recorded videos using Tracker software. For each of these experiments, the device is excited from zero initial conditions ensuring the sphere magnet starts from the same potential well. The sphere oscillates near one tethered position when excitation amplitude of 0.2 g is provided in figure 8(a), since the energy gained by the sphere magnet is not enough to conquer the potential barrier depth created by the tethering magnets. This kind of motion corresponds to output voltage shown in figure 8(a) or region A in figure 8(b). Chaotic cross-well motion is present in figure 9(b) where the energy gained is enough to occasionally travel across potential wells and rarely maintain revolutionary motion. Finally, at 0.8 g excitation the sphere magnet can take advantage of the full range of travelling path available as shown in figure 9(c), where periodic and continuous rotation is achieved. This motion is desirable since maximum change in magnetic flux occurs in the coils to provide optimized output power. Based on the intended application, the potential energy from the tethering magnets can be weakened by increasing the gap hrt to enable full range of motion while still continuing half-rotation of the sphere magnet between wells.

Figure 9.

Figure 9. Rolling motion time history from 1.9 Hz horizontal base excitation. Intra-well (a), chaotic cross-well (b) and full revolution (c) motion observed from the rolling magnet at 0.2 g, 0.4 g and 0.8 g input excitation amplitudes, respectively.

Standard image High-resolution image

The damping in the rolling path due to friction from the 3D printed casing is also measured using this camera setup. In figure 10(a), a frame of the Tracker software analysing the damped harmonic motion of the sphere magnet is shown and the extracted data are presented in figure 10(b). The peak values were used to calculate the damping coefficient as shown in equation (7):

Figure 10.

Figure 10. Damping measurement of the rolling magnet on the 3D printed casing surface using Tracker software (a) and the corresponding velocity-time history from this measurement (b).

Standard image High-resolution image

Equation (7)

where n is the number of velocity peaks in figure 10(b), $x(t_{0})$ and $x(t_{n})$ are the initial and final peak velocities and c is the mechanical damping coefficient.

4.4. Performance Analysis

In order to study the output performance of the proposed system across a broadband low frequency range, the theoretical model is used to predict the RMS output voltage and power achievable and to determine the boundary where the system can transition from a less desirable intra-well to high energy motion. A comparison between numerical model and experimental results is conducted to validate the model before exploring this output behaviour.

Figure 11 presents a comparison of the mean values between the experimental output and the calculated values of the numerical model in (a) and (b). The plot from 1.4 Hz excitations in figure 11(a) shows a very close match at 0.1 g and between 0.3 g–0.8 g amplitudes, though the theoretical model achieves continuous rotational motion at slightly smaller amplitude at 0.2 g compared to 0.3 g with its physical counterpart. The difference is caused due to a small mismatch in the excitation conditions in the experiment. The slider used to provide sinusoidal vibrations tend to contain more noise during low amplitude operation as shown in figure 11(c) relative to higher amplitude motion in figure 11(f). Figure 11(b) shows the comparison for 1.9 Hz excitation, where a close match can also be observed for intra-well and continuous rotational motion at 0.1 g–0.2 g and 0.5 g–0.8 g respectively. The difference at 0.3 g and 0.4 g is caused by the difference in excitation amplitude as shown in figure 11(f), along with other variations including uneven friction in the physical rolling path and the poloidal motion of the sphere magnet that contributes to the deviation from the model during the cross-well mode of operation. To compare the time domain behaviour of the system, voltage time histories at 0.2 g and 0.8 g amplitudes are presented for 1.9 Hz in figures 11(d)–(e) and (g)–(h) respectively. Intra-well mode occurs both in the theoretical (figure 11(d)) and experimental model (figures 11(e)), as 0.25 V peak AC voltage output with similar pattern is obtained due to the sphere magnet oscillating near one of the coils. During a higher excitation amplitude of 0.8 g, the sphere magnet rotates continuously in the looped circular path, and the large-amplitude AC voltage with comparable envelopes occur in figures 11(g) and (h) as the magnet travels past each coil. Hence, an overall agreement between experimental and numerical results can be concluded and the theoretical model is capable of describing the system dynamics.

Figure 11.

Figure 11. Close match between peak output voltage generated from theoretical model and experimental prototype at 1.4 Hz (a) and 1.9 Hz (b) excitation frequencies. At 1.9 Hz frequency, the input excitation waveforms (c) and (f), and voltage waveforms at 0.2 g (d)–(e) and 0.8 g (g)–(h) are compared.

Standard image High-resolution image

In order to examine the ideal operation conditions, a parametric investigation is conducted to define the critical limits for obtaining high output from the MEH. Figure 12(a) presents the RMS output voltage relationship with frequency and amplitude ranged between 0.5–5 Hz and 0.1–3 g respectively. The critical boundary is made apparent from the projected map in figure 12(b), where the minimum excitation amplitude for each frequency is shown for the system to transition from low-energy intra-well mode to desirable continuous rotation mode. The boundary also shows that for higher frequencies, it takes larger amplitudes to make this transition but the output power achieved in the desired region is consequently higher. Figures 12(c)–(e) present the phase portraits to gain further insights in the behaviour of the system at the blue region, near the boundary and the desired region respectively. The response is more chaotic near the boundary, as shown by the poincare map in figure 12(d) for 2 Hz and 0.45 g excitation, where 2.3 mW RMS power can be extracted. While stable response is achievable at the two regions, with intra-well response at 0.06 g 1.5 Hz to extract 0.3 mW in figure 12(c) and continuous rotation at 2 g 4 Hz excitation to get 4.1 mW as shown in figure 12(e).

Figure 12.

Figure 12. RMS voltage output of the energy harvester from an excitation range between 0.5 Hz–5 Hz and 0.1 g–3 g (a) and a projected map showing the RMS Power (b). Poincare map with phase portraits from three regions on the map: periodic intra-well motion in the blue region (c), cross-well motion near the boundary (d) and continuous rotation in the high output region (e).

Standard image High-resolution image

5. Self-Powered Sensing

The fabricated MEH is finally implemented as a part of a wireless sensor node to validate its use in a self-powered sensing application, and the operation is presented in figure 13. To design an intermediate energy storage system, the AC output from the energy harvester is converted to DC using a Schottky diode bridge rectifier and regulated by employing a BQ25505 evaluation board. The output voltage is then used to charge a 330 µF capacitor. The setup is illustrated in figure 13(a). The sensor node is programmed using a CC2650 low energy Bluetooth enabled microcontroller (MCU) module and an LMT85LP temperature sensor is connected it. The storage capacitor is charged to 3.3 V using the power management unit as shown in figure 13(b), where human motion near 2 Hz and 3 g excitation was harnessed and 1.91 mJ of energy is stored within 116 s with average power of 1.79 mW. After the capacitor charges to 3.3 V, the power management board triggers a signal to the MCU. The MCU is programmed with a reset mode such that it activates when triggered by the power management unit and it takes a measurement from the temperature sensor, and switches off when the capacitor discharges to 1.8 V. This charging and discharging cycle continue as long as the harvester is generating voltage, and the frequency of the measurements depends on the charging duration. The charging cycle is shown in figure 13(c) where the sensor is shown to take a reading every 10–15 s. The frequency of the sensor readings depends on the input excitation and can be increased by tuning the device for higher input energy available from activities like running.

Figure 13.

Figure 13. Self-powered sensing setup (a) consisting of the energy harvester, diode rectifier, power management board and a sensing unit. Storage capacitor is charged from human motion (b) and a temperature sensor is being powered as the capacitor is recharged between measurements (c).

Standard image High-resolution image

Table 2 provides a comparison of the MEH and the recent energy harvesting mechanisms established to extract energy from ambient vibrations, mostly focusing on rotational designs with electromagnetic transduction for human motion and low frequency oscillations. The bandwidth and the amplitude ranges used for comparison are derived numerically from the experimentally-validated theoretical model. The advantage of the proposed design with regard to output power generation, power density, and broadband low frequency range is thus made clear, along with the benefits established regarding multi-stability, bidirectional energy extraction, and optimised electromechanical coupling. It is also apparent that while the normalised density of the resonant energy harvester is slightly higher than that of this work, the bandwidth and the qualitative advantages from the mechanism are clear for ultra-low frequency broadband applications.

Table 2. Comparison with recently reported energy harvesting methods for low frequency vibrations.

ReferencesBandwidthAmplitudeMechanismVolume [cm3]Power [mW]Power Density [µW/cm3]Normalised Density [µW/cm3/g2/Hz]
Halim and Park [55] a 2.6–5 Hz2 gImpacts, rolling mass, frequency up-conversion230.1757.6 b 0.38
Zhang and Su [56]2–6 Hz0.5 gMagnetic spring36.80.2 c 5.43.6
Wang et al [57]9 km h−1 Magnetic attraction, rotary-translation40.60.297.1
Halim et al [58]5.6 km h−1 ±25 swingeccentric pendulum, sprung rotor20.10.063
Smilek et al [59]2–10 Hz0.2–1 gTusi couple501.4 d 20.136.4
Li et al [24]20 Hz1 gResonant, abrupt flux density change37.42841100 e 55
This work0.5–5 Hz0.1–3 gMulti-stable nonlinearity, magnetic tethering, bidirectional1405.78 f 41.4 g 46.2

a Piezoelectric transduction employed here. The rest are electromagnetic transduction. b 4.96 Hz excitation. c 6 Hz and 0.5 g excitation. d 3.45 Hz and 0.4 g excitation. e Value given in [24]. f Based on the conditions used in figure 6. g The volume of the casing along with the coil and tethering magnet array are considered.

6. Conclusions

This paper introduces a novel electromagnetic nonlinear energy harvester designed to address challenges in power generation from random and broadband low-frequency vibration sources for real-time monitoring applications. By incorporating multi-stability in a sphere magnet's circular travelling path, the proposed energy harvester effectively converts bi-directional low-frequency excitation into rotational motion, allowing effective mechanical energy harnessing. Multi-stability is realised by placing eight tethering magnets in the rolling path to allow the proposed system to operate over a broad bandwidth and perform rolling without slipping movement to optimise rate of change of magnetic flux in the coils mounted to convert mechanical energy to electricity. A theoretical model is developed to study the electromechanical properties and unique dynamics of the multi-stable rotational energy harvester. Three distinct modes of operation, namely intra-well, cross-well, and continuous rotational motion are identified from the theoretical study, which are dependent on the level of excitation available from the sources. Low amplitude motion results in intra-well mode of operation, while continuous rotation is achieved at higher amplitudes where periodic regime behaviour occurs and is desired for power generation. A physical prototype is fabricated and validated using the theoretical model, and is employed as a part of a self-powered sensing system. The device can achieve a maximum power of 5.8 mW is achievable by the device with excitation values as low as 1.4 Hz and 0.8 g. The harvester can operate in continuous periodic rotational motion between 1.4 and 1.9 Hz with excitation amplitude as low as 0.3 g. The numerical investigation explores its operating range further by establishing a relationship between power generation for excitation ranged from 0.1 g to 3 g and 0.5 Hz–5 Hz. The study shows that the desired high output operation is separated by a boundary, below which low-energy intra-well motion occurs. For higher frequencies, higher amplitude excitation is required to transition to the desired mode of operation, and RMS power as high as 5.1 mW is achievable with 5 Hz and 3 g excitation. This unique energy harvesting solution opens new avenues for power autonomous sensing from ultra-low frequency broadband vibration environments with bidirectional harnessing feature, having the potential to address the energy needs of low-power electronic devices in remote wireless sensing systems. Future work includes manufacturing an optimised iteration of this design to reduce losses due to friction and misalignments, integrate the storage and computing modules into the harvester, and scale the device for testing with dedicated low-frequency applications.

Acknowledgments

The authors would like to thank Wolfson School of Mechanical, Electrical and Manufacturing Engineering, Loughborough University for financially supporting this work.

Data availability statement

The data cannot be made publicly available upon publication because they contain commercially sensitive information. The data that support the findings of this study are available upon reasonable request from the authors.

Conflict of interest

The authors declare that there are no known conflicts of interest to influence the work reported in this paper.

Please wait… references are loading.