Convergence acceleration for partitioned simulations of the fluid-structure interaction in arteries

Abstract

We present a partitioned approach to fluid-structure interaction problems arising in analyses of blood flow in arteries. Several strategies to accelerate the convergence of the fixed-point iteration resulting from the coupling of the fluid and the structural sub-problem are investigated. The Aitken relaxation and variants of the interface quasi-Newton -least-squares method are applied to different test cases. A hybrid variant of two well-known variants of the interface quasi-Newton-least-squares method is found to perform best. The test cases cover the typical boundary value problem faced when simulating the fluid-structure interaction in arteries, including a strong added mass effect and a wet surface which accounts for a large part of the overall surface of each sub-problem. A rubber-like Neo Hookean material model and a soft-tissue-like Holzapfel-Gasser-Ogden material model are used to describe the artery wall and are compared in terms of stability and computational expenses. To avoid any kind of locking, high-order finite elements are used to discretize the structural sub-problem. The finite volume method is employed to discretize the fluid sub-problem. We investigate the influence of mass-proportional damping and the material model chosen for the artery on the performance and stability of the acceleration strategies as well as on the simulation results. To show the applicability of the partitioned approach to clinical relevant studies, the hemodynamics in a pathologically deformed artery are investigated, taking the findings of the test case simulations into account.