Adaptive, second-order, unconditionally stable partitioned method for fluid–structure interaction

We propose a novel, time adaptive, strongly-coupled partitioned method for the interaction between a viscous, incompressible fluid and a thin elastic structure. The time integration is based on the refactorized Cauchy’s one-legged ‘θ−like’ method, which consists of a backward Euler method using a θτ...

Full description

Saved in:
Bibliographic Details
Published in:Computer methods in applied mechanics and engineering Vol. 393; p. 114847
Main Authors: Bukač, Martina, Trenchea, Catalin
Format: Journal Article
Language:English
Published: Amsterdam Elsevier B.V 01.04.2022
Elsevier BV
Subjects:
ISSN:0045-7825, 1879-2138
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We propose a novel, time adaptive, strongly-coupled partitioned method for the interaction between a viscous, incompressible fluid and a thin elastic structure. The time integration is based on the refactorized Cauchy’s one-legged ‘θ−like’ method, which consists of a backward Euler method using a θτn-time step and a forward Euler method using a (1−θ)τn-time step. The bulk of the computation is done by the backward Euler method, as the forward Euler step is equivalent to (and implemented as) a linear extrapolation. The variable τn-time step integration scheme is combined with the partitioned, kinematically coupled β−scheme, used to decouple the fluid and structure sub-problems. In the backward Euler step, the two sub-problems are solved in a partitioned sequential manner, and iterated until convergence. Then, the fluid and structure sub-problems are post-processed/extrapolated in the forward Euler step, and finally the τn-time step is adapted. The refactorized Cauchy’s one-legged ‘θ−like’ method used in the development of the proposed method is equivalent to the midpoint rule when θ=12, in which case the method is non-dissipative and second-order accurate. We prove that the sub-iterative process of our algorithm is linearly convergent, and that the method is unconditionally stable when θ≥12. The numerical examples explore the properties of the method when both fixed and variable time steps are used, and in both cases shown an excellent agreement with the reference solution.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0045-7825
1879-2138
DOI:10.1016/j.cma.2022.114847