Global format for energy-momentum based time integration in nonlinear dynamics

    Research output: Contribution to journalJournal articleResearchpeer-review

    588 Downloads (Pure)

    Abstract

    A global format is developed for momentum and energy consistent time integration of second‐order dynamic systems with general nonlinear stiffness. The algorithm is formulated by integrating the state‐space equations of motion over the time increment. The internal force is first represented in fourth‐order form consisting of the end‐point mean value plus a term containing the stiffness matrix increment. This form gives energy conservation for systems with internal energy as a quartic function of the displacement components. This representation is then extended to general energy conservation via a discrete gradient representation. The present procedure works directly with the internal force and the stiffness matrix at the time integration interval end‐points, and in contrast to previous energy‐conserving algorithms, it does not require any special form of the energy function nor use of mean value products at the element level or explicit use of a geometric stiffness matrix. An optional monotonic algorithmic damping, increasing with response frequency, is developed in terms of a single damping parameter. In the solution procedure, the velocity is eliminated and the nonlinear iterations are based on the displacement components alone. The procedure represents an energy consistent alternative to available collocation methods, with an equally simple implementation. Copyright © 2014 John Wiley & Sons, Ltd.
    Original languageEnglish
    JournalInternational Journal for Numerical Methods in Engineering
    Volume100
    Issue number6
    Pages (from-to)458-476
    Number of pages19
    ISSN0029-5981
    DOIs
    Publication statusPublished - 2014

    Keywords

    • Time integration
    • Nonlinear dynamics
    • Energy conservation

    Fingerprint

    Dive into the research topics of 'Global format for energy-momentum based time integration in nonlinear dynamics'. Together they form a unique fingerprint.

    Cite this