Experimental evaluation of Tusi couple based energy harvester for scavenging power from human motion

This paper deals with the experimental performance evaluation of the prototype of a novel inertial energy harvester based on Tusi couple mechanism. The harvester was developed as an autonomous power source for environments with very low frequency and magnitude of mechanical vibrations available. The experiments were conducted using human body during different activities as a source of mechanical excitation, with the prospect of using the harvester for powering up future wearable electronic devices. Four different locations on a single measurement specimen were picked for the harvester placement – back of the head, belt, wrist and ankle. Measurements in each location comprised of walking on a straight and level path at natural speed, walking up and down the stairs, jumping, running, and location-specific activities that were expected to provide significant output power. The measured average output power of the device with dimensions 50x50x20 mm on empirically selected 2 kΩ electrical load reached up to 6.5 mW, obtained with the device attached to the ankle while shaking the leg.


Introduction
Converting the energy of human motion into electricity has been gaining a significant interest in past few decades thanks to its obvious exploitability in biomedical and low power wearable electronic applications.Various designs of energy harvesters have been presented, utilizing mostly piezoelectric effect [1,2]; electromagnetic induction [3,4]; or hybrid approaches [5] to convert the mechanical energy into electricity.Recent works also report utilizing a triboelectric effect [6] or electrochemically driven harvesters [7] for the same goal.The size of devices ranges from large devices integrated into a backpacks or special braces [8,9], through harvesters small enough to be implemented into apparel or wearable devices [10], to implantable MEMS energy harvesters, meant for powering up biomedical implants [11].The harvester presented in this paper with its dimensions 50x50x20mm falls into the larger devices category, attachable externally to various parts of human body, or into dedicated pockets in smart clothing products, as close to the intended application as possible.https://doi.org/10.1051/matecconf/201821105004VETOMAC XIV

Harvester design
The harvester (Fig. 1) comprises of a rolling proof mass with 6 permanent magnets fixed to each face of the rolling mass.Movement of the proof mass with 1 dof in the circular cavity of double the diameter of the proof mass causes the magnets to pass in front of the pick-up coils, placed above the centre of the cavity.The neighbouring magnets are fixed to the rolling mass with alternating magnetization direction to maximize the magnetic flux change during the motion.The movement of the proof mass is described by the Lagrange equation of second kind, using q as generalized coordinate: d/dt(dEk/dq̇) -dEk/dq + dEb/dq̇ + dEp/dq = -dA/dq = -Q (1) After a series of mathematical operations the final form of this equation, valid for this design of the harvester where the proof mass centre of gravity is aligned with its geometrical centre, can be derived as: Where Itotal the moment of inertia of the proof mass with respect to instantaneous axis of rotation, bm * denotes mechanical damping, m is total weight of the proof mass, R is radius of the cavity, and r is radius of the proof mass disc.z̈x and z̈y denote excitation acceleration in two perpendicular directions in the working plane of the harvester.The useful electrical energy is dissipated on electric load RL, which contributes to the electrical damping of the system be * , together with the resistance of the pick-up coils connected in series RC: The RMS value of magnetic flux change with respect to the displacement of the harvester dΦ/dq reaches 0.2 Wb/rad for the evaluated magnetic circuit design.

Experimental setup and methodology
The experiment was conducted with the harvester device fixed on a testing rig together with a three-axial accelerometer to log the acceleration in both working axes of the harvester.A single test subject (male, 183 cm, 75 kg) was then performing a series of activities with the harvester mounted on different locations on the body, which were empirically deemed feasible for potential harvester placement (Fig. 2).The locations selected were back of the head, belt, wrist and ankle.Each of the measurement runs lasted 300 s, with six to seven different actions being performed during two runs in each harvester placement.The first measurement in each placement consisted solely of the subject walking at natural speed on level surface inside a building, second measurement contained placement-specific activities.The recorded activities common for all the harvester placements were running, going up and down the stairs, and jumping.The location specific activities included nodding and shaking the head, jumping jacks, different walking style patterns or shaking the limbs violently.The harvested power was measured on 2 kΩ electrical load, and recorded using NI-9234 card.Since the peaks of generated voltage on load reached beyond the card ±5 V limits, a voltage divider was employed and the recorded voltages were then recalculated to obtain the power dissipated on the full load.

Results
The acceleration measurements were exploited to identify the different recorded activities in the second set of measurements (Fig. 3).Furthermore, the FFT was calculated for each activity and placement, and the dominant acceleration peak in both working directions of the harvester was noted for future analyses.The recorded voltage waveforms on the 2 kΩ electrical load (Fig. 4) were used to calculate the RMS values of voltage and average electrical power dissipated on load (Tab.1).https://doi.org/10.1051/matecconf/201821105004VETOMAC XIV Fig. 3. Voltage on load and acceleration in the relevant axes with harvester placement on head during the second measurement set.The results demonstrate, that depending on the placement, this harvester prototype can be used to power up some of the low-power wearable or implantable electronic applications, at least during the time of the user activity.Taking into account the 0.87 hours of average time per day that is spent by walking or exploitable activities [12], the average power available through the whole 24 hours of the day ranges between 2 μW for the head placement (worst case) and 52 μW for the ankle placement (best case).

Conclusions
The paper presented an experimental performance evaluation of a novel design of electromagnetic energy harvester developed for human power harvesting.The harvester was tested in real-life conditions, mounted on different locations on testing subject, performing different activities ranging from walking to violent limbs shaking.The harvester was attached in the measurements locations together with a 3-axis acceleration datalogger in order to record the input acceleration data for possible future design MATEC Web of Conferences 211, 05004 (2018) https://doi.org/10.1051/matecconf/201821105004VETOMAC XIV optimization.Average power on load harvested during the normal walking varied between 56 μW for the harvester placed on the back on the head, and 1.4 mW for the device attached on the ankle.The results indicate that, assuming a reasonably active behaviour of the user, the device could provide a feasible alternative power source for modern low-power health sensors and wearable electronics; such as wristwatches, temperature or humidity sensors, accelerometers or pressure sensors.Furthermore, optimization of the harvester design for the conditions in particular locations could increase the output power levels beyond the current results.

Fig. 2 .
Fig. 2. Measurement setup and the harvester with datalogger in three out of four measured locations.

Fig. 4 .
Fig. 4. Generated voltage on load during normal walking in all measured harvester placements.

Table 1 .
org/10.1051/matecconf/201821105004VETOMAC XIV RMS values of voltage and average power on load for different placements and activities.