A novel class of highly efficient and accurate time-integrators in nonlinear computational mechanics

Xuechuan Wang, Satya N. Atluri

Research output: Contribution to journalArticlepeer-review

9 Scopus citations

Abstract

A new class of time-integrators is presented for strongly nonlinear dynamical systems. These algorithms are far superior to the currently common time integrators in computational efficiency and accuracy. These three algorithms are based on a local variational iteration method applied over a finite interval of time. By using Chebyshev polynomials as trial functions and Dirac–Delta functions as the test functions over the finite time interval, the three algorithms are developed into three different discrete time-integrators through the collocation method. These time integrators are labeled as Chebyshev local iterative collocation methods. Through examples of the forced Duffing oscillator, the Lorenz system, and the multiple coupled Duffing equations (which arise as semi-discrete equations for beams, plates and shells undergoing large deformations), it is shown that the new algorithms are far superior to the 4th order Runge–Kutta and ODE45 of MATLAB, in predicting the chaotic responses of strongly nonlinear dynamical systems.

Original languageEnglish
Pages (from-to)861-876
Number of pages16
JournalComputational Mechanics
Volume59
Issue number5
DOIs
StatePublished - May 1 2017

Keywords

  • Chaos
  • Chebyshev polynomials
  • Collocation method
  • Strongly nonlinear system
  • Variational iteration method

Fingerprint Dive into the research topics of 'A novel class of highly efficient and accurate time-integrators in nonlinear computational mechanics'. Together they form a unique fingerprint.

Cite this