Generalized convolution quadrature based boundary element method for uncoupled thermoelasticity
•Application of the generalised Convolution Quadrature in thermoelastic BEM.•Variable time step size in a time domain thermoelastic BE formulation.•Convergence study of a CQ based BEM.•Realisation of Robin boundary condition for different shape functions of the Dirichlet and Neumann data. Mechanical...
Gespeichert in:
Veröffentlicht in: | Mechanical systems and signal processing 2021-03, Vol.150, p.107234, Article 107234 |
---|---|
Hauptverfasser: | , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | •Application of the generalised Convolution Quadrature in thermoelastic BEM.•Variable time step size in a time domain thermoelastic BE formulation.•Convergence study of a CQ based BEM.•Realisation of Robin boundary condition for different shape functions of the Dirichlet and Neumann data.
Mechanical loads together with changing temperature conditions can be found in a wide variety of fields. Their effects on elastic media are reflected in the theory of thermoelasticity. For typical materials in engineering, very often a simplification of this coupled theory can be used, the so-called uncoupled quasistatic thermoelasticity. Therein, the effects of the deformations onto the temperature distribution is neglected and the mechanical inertia effects as well. The Boundary Element Method is used to solve numerically these equations in three dimensions. Since convolution integrals occur in this boundary element formulation, the Convolution Quadrature Method may be applied. However, very often in thermoelasticity the solution shows rapid changes and later on very small changes. Hence, a time discretisation with a variable time step size is preferable. Therefore, here, the so-called generalised Convolution Quadrature is applied, which allows for non-uniform time steps. Numerical results show that the proposed method works. The convergence behavior is, as expected, governed either by the time stepping method or the spatial discretisation, depending on which rate is smaller. Further, it is shown that for some problems the proposed use of the generalised Convolution Quadrature is the preferable. |
---|---|
ISSN: | 0888-3270 1096-1216 |
DOI: | 10.1016/j.ymssp.2020.107234 |