A modular,operator‐splitting scheme for fluid–structure interaction problems with thick structures

We present an operator‐splitting scheme for fluid–structure interaction (FSI) problems in hemodynamics, where the thickness of the structural wall is comparable to the radius of the cylindrical fluid domain. The equations of linear elasticity are used to model the structure, while the Navier–Stokes equations for an incompressible viscous fluid are used to model the fluid. The operator‐splitting scheme, based on the Lie splitting, separates the elastodynamics structure problem from a fluid problem in which structure inertia is included to achieve unconditional stability. We prove energy estimates associated with unconditional stability of this modular scheme for the full nonlinear FSI problem defined on a moving domain, without requiring any sub‐iterations within time steps. Two numerical examples are presented, showing excellent agreement with the results of monolithic schemes. First‐order convergence in time is shown numerically. Modularity, unconditional stability without temporal sub‐iterations, and simple implementation are the features that make this operator‐splitting scheme particularly appealing for multi‐physics problems involving FSI. Copyright © 2013 John Wiley & Sons, Ltd.     

A fictitious domain/distributed Lagrange multiplier method for the particulate flow of Oldroyd-B fluids: A positive definiteness preserving approach

In this article we present a numerical method for simulating the sedimentation of circular particles in a two-dimensional channel filled with an Oldroyd-B fluid. We have combined a fictitious domain/distributed Lagrange multiplier method with a factorization approach from Lozinski and Owens [J. Non-Newtonian Fluid Mech. 112 (2003) 161] via an operator splitting technique. The new scheme preserves the positive definiteness of the conformation tensor at the discrete level. The method is validated by performing a convergence study which shows that the results are independent of the mesh and time step sizes. Our results show that when the elasticity number (E) is less than a critical value (which depends upon the blockage ratio), two particles will sediment in the channel-like particles in Newtonian fluids; when the elasticity number is greater than the critical value, chains are formed for the case of two particles sedimenting in an Oldroyd-B fluid and the center line is aligned with the falling direction. These results agree with those presented in [P.Y. Huang, H.H. Hu, and D.D. Joseph, J. Fluid Mech. 362 (1998) 297]. For the cases of three and six particles, when the elasticity number is greater than a critical value and the viscoelastic Mach number is less than one, chains are also formed and move to the center of the channel.     

A boundary multiplier/fictitious domain method for the steady incompressible Navier-Stokes equations

Summary. We analyze the error of a fictitious-domain method with boundary Lagrange multiplier. It is applied to solve a non-homogeneous steady incompressible Navier-Stokes problem in a domain with a multiply-connected boundary. The interior mesh in the fictitious domain and the boundary mesh are independent, up to a mesh-length ratio. Received February 24, 1999 / Revised version received January 30, 2000 / Published online October 16, 2000     

On an Inequality of C. Sundberg: A Computational Investigation via Nonlinear Programming

The main goal of this article is to discuss a numerical method for finding the best constant in a Sobolev type inequality considered by C. Sundberg, and originating from Operator Theory. To simplify the investigation, we reduce the original problem to a parameterized family of simpler problems, which are constrained optimization problems from Calculus of Variations. To decouple the various differential operators and nonlinearities occurring in these constrained optimization problems, we introduce an appropriate augmented Lagrangian functional, whose saddle-points provide the solutions we are looking for. To compute these saddle-points, we use an Uzawa–Douglas–Rachford algorithm, which, combined with a finite difference approximation, leads to numerical results suggesting that the best constant is about five times smaller than the constant provided by an analytical investigation.     

Symmetry breaking and preliminary results about a Hopf bifurcation for incompressible viscous flow in an expansion channel

This computational study shows, for the first time, a clear transition to two-dimensional Hopf bifurcation for laminar incompressible flows in symmetric plane expansion channels. Due to the well-known extreme sensitivity of this study on computational mesh, the critical Reynolds numbers for both the known symmetry-breaking (pitchfork) bifurcation and Hopf bifurcation were investigated for several layers of mesh refinement. It is found that under-refined meshes lead to an overestimation of the critical Reynolds number for the symmetry breaking and an underestimation of the critical Reynolds number for the Hopf bifurcation.     

Numerical study of the wall effect on particle sedimentation

ABSTRACT

In this article, we investigate the abnormal settling of two-disk systems and elliptical shaped particles in infinite two-dimensional channels filled with an incompressible viscous fluid. We apply a distributed Lagrange multiplier/fictitious domain method (DLM/FDM) for the direct numerical simulation of these particulate flows. Due to the wall effect, the two-disk systems can form chains which settle stably instead of having the particles moving apart. Also, sedimentation with the long axis moving to vertical positions in the middle of the infinite channel has been observed for the elliptic shaped particles. The critical Reynolds number for having such an abnormal settling behaviour decreases as the width of the channel increases.     

Tuning the mesh of a mixed method for the stream function Vorticity formulation of the Navier-Stokes equations

Summary In this article we study a new mixed method for the Stokes and Navier-Stokes equations. The method uses two meshes, one very fine for and a coarser one for . Error estimates show that boundary layers do not require to refine the mesh for the stream function as much as for the vorticity when the Reynolds number is large. We prove estimates and study implementation problems.