Eventos Anais de eventos
COBEM 2021
26th International Congress of Mechanical Engineering
ROBUST DESIGN OF ENERGY HARVESTING RESONANT DEVICES BY MULTI-OBJECTIVE OPTIMIZATION TECHNIQUES AND POLYNOMIAL CHAOS EXPANSIONS
Submission Author:
Paulo Martins , SP
Co-Authors:
Paulo Martins, Marcelo A. Trindade, paulo varoto
Presenter: Paulo Martins
doi://10.26678/ABCM.COBEM2021.COB2021-0716
Abstract
In recent years, energy harvesting from mechanical vibrations has been a very active research topic. This can be achieved with help of piezoelectric materials which can be used to convert mechanical energy into electricity. This is particularly interesting to energize moving or portable devices, which still rely heavily on batteries. Nevertheless, the amount of energy available for conversion is usually small and the conversion efficiency is highly sensitive to uncertainties found in the harvesting devices themselves and their environment. Thus, the design of energy harvesting devices should account for uncertainty analysis and robust optimization techniques. This work presents some recent results on the robust design of energy harvesting devices, based on cantilever resonating beams, when subjected to uncertainties on the beam clamping, harvesting circuit and effective damping. Multi-objective optimization techniques are considered to search for optimal values of design parameters, namely beam length, piezoeletric patch length, position and thickness and circuit effective resistance, leading to satisfactory compromises between mean and dispersion of the potentially harvested energy. Due to the high computational cost of uncertainty quantification, estimation of mean and dispersion of the response is found using Polynomial Chaos Expansions (PCE) and two multi-objective techniques are considered, Compromise Programming (CP) and Non-dominated Sorting Genetic Algorithm (NSGA-II). Results show that harvesting devices with smaller length and larger masses generally lead to best nominal performance but also to higher dispersions. Also, the dispersions can be reduced by using effective circuit resistances smaller than the nominal values. Several design alternatives are presented in the form of a Pareto front, from which a designer can choose based on the desired compromise.
Keywords
energy harvesting, piezoelectric sensors, Uncertainties, robust project, multiobjective optimization
