Invariant manifolds and the parameterization method in coupled energy harvesting piezoelectric oscillators

Albert Granados

Research output: Contribution to journalJournal articleResearchpeer-review


Energy harvesting systems based on oscillators aim to capture energy from mechanical oscillations and convert it into electrical energy. Widely extended are those based on piezoelectric materials, whose dynamics are Hamiltonian submitted to different sources of dissipation: damping and coupling. These dissipations bring the system to low energy regimes, which is not desired in long term as it diminishes the absorbed energy. To avoid or to minimize such situations, we propose that the coupling of two oscillators could benefit from theory of Arnold diffusion. Such phenomenon studies O(1) energy variations in Hamiltonian systems and hence could be very useful in energy harvesting applications. This article is a first step towards this goal. We consider two piezoelectric beams submitted to a small forcing and coupled through an electric circuit. By considering the coupling, damping and forcing as perturbations, we prove that the unperturbed system possesses a 4-dimensional Normally Hyperbolic Invariant Manifold with 5 and 4-dimensional stable and unstable manifolds, respectively. These are locally unique after the perturbation. By means of the parameterization method, we numerically compute parameterizations of the perturbed manifold, its stable and unstable manifolds and study its inner dynamics. We show evidence of homoclinic connections when the perturbation is switched on.
Original languageEnglish
JournalPhysica D: Nonlinear Phenomena
Pages (from-to)14-29
Publication statusPublished - 2017


  • Damped oscillators
  • Energy harvesting systems
  • Parameterization method
  • Normally hyperbolic invariant manifolds
  • Homoclinic connections
  • Arnold diffusion


Dive into the research topics of 'Invariant manifolds and the parameterization method in coupled energy harvesting piezoelectric oscillators'. Together they form a unique fingerprint.

Cite this