Energy Harvesters (EH) are devices that convert environmental energy (i.e. thermal, vibrational, solar or electromagnetic) into electrical energy. One of the most promising solutions consists in transforming energy from vibrations using a piezoelectric material placed onto a mechanical resonator. The intrinsic drawback of this solution is the typically high quality factor of the device and so the device works effectively only within a narrow bandwidth. To overcome this limitation it is possible to tune the mechanical resonance of the device, to introduce non-linear elements (e.g. magnets) or to design the mechanical resonator with a multimodal behavior. In Ultra Low Power (ULP) applications the aspect of integration is of utmost importance and so MEMS-based (micro electro-mechanical systems) EHs are preferable. Within this scenario the multimodal solution is the more suitable considering the technological constraints imposed by the micro machining manufacturing process.
In this paper we optimize a given multimodal mechanical geometry in order to maximize the number of resonances within a certain frequency band. In the particular case of piezoelectric energy harvesting, the strain distribution of each modes is critical and has to be taken into consideration for the designing of efficient device. The proposed optimization is FEM-based and it uses modal and harmonic simulations for both select the useful modes and then to design the device in a way that presents those modes within a predefined frequency range. This mechanical optimization could be considered the first step for maximizing the output power of a multimodal piezoelectric energy harvester. The second step focuses on the geometry optimization of the piezoelectric transducer element, starting from the desired resonant mode configuration defined in the first stage. The number of modes stimulated applying a vertical acceleration increases in number in the desired frequency range (i.e. around 1 kHz). As the output power is proportional to the stress of the flexible device, these promising results show clearly how an optimization of the geometry could significantly boost the performance of such devices.
