A stable partitioned FSI algorithm for rigid bodies and incompressible flow in three dimensions

Qi Tang, Jeffrey W. Banks, William D. Henshaw, Donald W. Schwendeman, Qi Tang 0001

Abstract

Simulating the interaction between fluids and solids (FSI) is a cornerstone of computational engineering, yet it remains plagued by numerical instabilities when solids are light relative to the surrounding fluid. This paper introduces the first "partitioned" algorithm for FSI that maintains stability and second-order accuracy in three dimensions without the need for computationally expensive sub-time-step iterations. The "Added-Mass Partitioned" (AMP) scheme successfully models the motion of extremely light—and even zero-mass—rigid bodies in incompressible flows, addressing the "added-mass" effect that typically causes traditional partitioned algorithms to fail.

Introduction: The Added-Mass Challenge

In fluid-structure interaction simulations, researchers typically choose between monolithic schemes (stable but complex) and partitioned schemes (flexible but often unstable). The primary source of instability in partitioned schemes is the "added-mass" effect, which occurs when the fluid's inertia acts as an additional mass on a light solid, such as a heart valve in blood. Traditional fixes involve performing multiple sub-iterations within a single time step to synchronize the fluid and solid, but this approach is computationally prohibitive for large-scale 3D simulations. This work extends the Added-Mass Partitioned (AMP) framework to handle the full complexity of 3D rigid body motion, including arbitrary rotations.

Methodology: The AMP-RB Algorithm

The authors propose the AMP-RB (Added-Mass Partitioned - Rigid Body) algorithm. This scheme is built upon several key innovations designed to decouple the fluid and solid solvers while maintaining physical consistency:

  • Advanced Interface Conditions: Instead of merely matching velocities at the boundary, the AMP scheme derives a novel pressure boundary condition. This condition balances the translational and rotational accelerations of the solid with the fluid's pressure and viscous stresses.
  • Added-Damping Tensors: To counteract secondary instabilities arising from viscous fluid stresses, the authors generalized "added-damping" mathematics into four $3 \times 3$ tensors. These tensors are essential for handling the non-linear nature of 3D rotations.
  • Moving Overlapping Grids: Utilizing the Overture framework, the algorithm employs structured grids that overlap and move in tandem with the rigid bodies, ensuring high mesh quality around complex geometries.
Definitions of the kinematic and inertial variables used to describe 3D rigid body motion within the AMP-RB framework.

Validation and Benchmarking

The algorithm was rigorously validated against several benchmarks to ensure accuracy and stability:

  1. 1D Piston: Verified against analytical solutions for exact accuracy.
  2. Settling Sphere: Compared with experimental data of a sphere dropping in silicon oil.
  3. Low-Density Particle: A specific test for stability at extreme solid-to-fluid density ratios.
  4. Bi-leaflet Mechanical Heart Valve: A complex 3D benchmark involving hinged leaflets in a pulsating flow, demonstrating the method's applicability to medical engineering.
Simulation of a bi-leaflet mechanical heart valve in both open and closed states, showing the complex flow patterns and the overlapping grid system used for the leaflets.

Key Results and Insights

  • The "Massless" Breakthrough: A significant finding is that the algorithm remains stable even when the rigid body has zero mass. In this scenario, the body's motion is governed entirely by the fluid's added mass and damping.
  • Efficiency and Accuracy: The AMP scheme proved to be 1.65x to 1.73x faster than traditional partitioned schemes by avoiding sub-iterations. Furthermore, it demonstrated strict second-order accuracy in both time and space across all tested densities.
  • Parallel Performance: Implemented using MPI and the PETSc library, the code showed effective strong scaling on distributed-memory clusters, though pressure linear solvers remain a primary computational bottleneck.

Conclusion and Future Work

The AMP-RB algorithm provides a robust and efficient path forward for 3D FSI simulations involving light rigid bodies. While the current results are promising, future research will focus on developing specialized multigrid solvers to improve the conditioning of the pressure linear systems and applying the algorithm to higher Reynolds number turbulent flows and more anatomically realistic heart valve geometries.

Summary updated

Abstract

This paper describes a novel partitioned algorithm for fluid-structure interaction (FSI) problems that couples the motion of rigid bodies and incompressible flow. This is the first partitioned algorithm that remains stable and second-order accurate, without sub-time-step iterations, for very light, and even zero-mass, bodies in three dimensions. This new added-mass partitioned (AMP) algorithm extends the previous developments in [1, 2] by generalizing the added-damping tensors to account for arbitrary three-dimensional rotations, and by employing a general quadrature for the surface integral over a rigid body to derive the discrete AMP interface condition for the fluid pressure. Stability analyses for two three-dimensional model problems show that the algorithm remains stable for bodies of any mass when applied to the relevant model problems. The resulting AMP algorithm is implemented in parallel using a moving composite grid framework to treat one or more rigid bodies in complex three-dimensional configurations. The new three-dimensional algorithm is verified and validated though several benchmark problems, including the motion of a sphere in a viscous incompressible fluid and the interaction of a bi-leaflet mechanical heart valve and a pulsating fluid. Numerical simulations confirm the predictions of the stability analysis even for complex problems, and show that the AMP algorithm remains stable, without sub-iterations, for light and even zero-mass three-dimensional rigid bodies of general shape. These benchmark problems are further used to examine the parallel performance of the algorithm and to investigate the conditioning of the linear system for the pressure including the newly derived AMP interface conditions.
BibTeX
@article{tang2018stable,
  author = {Qi Tang and Jeffrey W. Banks and William D. Henshaw and Donald W. Schwendeman and Qi Tang},
  title = {{A stable partitioned FSI algorithm for rigid bodies and incompressible flow in three dimensions}},
  journal = {Journal of Computational Physics},
  year = {2018},
  doi = {10.1016/j.jcp.2018.06.072},
}
Figure