Performance limit for base-excited energy harvesting, and comparison with experiments

Abstract

We consider the theoretical maximum extractable average power from an energy harvesting device attached to a vibrating table which provides a unidirectional displacement $A\sin(\omega t)$. The total mass of moving components in the device is $m$. The device is assembled in a container of dimension $L$, limiting the displacements and deformations of components within. The masses in the device may be interconnected in arbitrary ways. The maximum extractable average power is bounded by $\frac{mLA\omega^3}{\pi}$, for motions in 1, 2, or 3 dimensions; with both rectilinear and rotary motions as special cases; and with either single or multiple degrees of freedom. The limiting displacement profile of the moving masses for extracting maximum power is discontinuous, and not physically realizable. But smooth approximations can be nearly as good: with $15$ terms in a Fourier approximation, the upper limit is $99$\% of the theoretical maximum. Purely sinusoidal solutions are limited to $\frac{\pi}{4}$ times the theoretical maximum. For both single-degree-of-freedom linear resonant devices and nonresonant whirling devices where the energy extraction mimics a linear torsional damper, the maximum average power output is $\frac{mLA\omega^3}{4}$. Thirty-six experimental energy harvesting devices in the literature are found to extract power amounts ranging from $0.0036$\% to $29$\% of the theoretical maximum. Of these thirty-six, twenty achieve less than 2\% and three achieve more than 20\%. We suggest, as a figure of merit, that energy extraction above $\frac{0.2 mLA\omega^3}{\pi}$ may be considered excellent, and extraction above $\frac{0.3 mLA\omega^3}{\pi}$ may be considered challenging.