A comparative study of time-marching schemes for fluid-structure interactions (original) (raw)
Related papers
A new staggered scheme for fluid-structure interaction
International Journal for Numerical Methods in Engineering, 2013
Staggered solution procedures represent the most elementary computational strategy for the simulation of fluid-structure interaction problems. They usually consist of a predictor followed by the separate execution of each subdomain solver. Although it is generally possible to maintain the desired order of accuracy of the time integration, it is difficult to guarantee the stability of the overall computation. In the context of large solid over fluid mass ratios, compressible flows and explicit subsolvers, substantial development has been carried out by Felippa, Park, Farhat, Löhner and others. In this work, a new staggered scheme is presented. It is shown that, for a linear model problem, the scheme is second-order accurate and unconditionally stable. The dependency of the leading truncation error on the solid over fluid mass ratio is investigated. The strategy is applied to two-dimensional and three-dimensional fluid-structure interaction problems. It is shown that the conclusions derived from the investigation of the model problem apply. The new strategy extends the applicability of staggered schemes to problems involving relatively small solid over fluid mass ratios and incompressible fluid flow. It is suggested that the proposed scheme has the same range of applicability as the Dirichlet-Neumann or block Gauß-Seidel type strategies. be computationally more efficient than strongly coupled solution procedures because they do not require iteration within each time step. For many problems, the computational time is one order of magnitude shorter for staggered than for strongly coupled strategies. Furthermore, staggered schemes allow for the employment of existing highly specialised subsolver software, which is generally not the case for the strongly coupled Newton-Raphson procedures. The drawback of staggered schemes is their restricted range of applicability as described next. Substantial research and development in the area of staggered schemes have been carried out by Felippa, Park, Farhat, Piperno, Lesoinne, Löhner and co-workers, see, for example, .
Time marching for simulation of fluid–structure interaction problems
Journal of Fluids and Structures, 2009
Numerical simulation of industrial multi-physics problems is still a challenge. It generally requires large computational resources. It may involve complex code coupling techniques. It also relies on appropriate numerical methods making data transfer possible, quick and accurate. In the framework of partitioned procedures, multi-physics computations require the right choice of code coupling schemes, because several physical mechanisms are involved. Numerical simulation of fluid-structure interactions is one of these issues. It is investigated in this paper. First the computational process involving a code coupling procedure is presented. Then, applications and test cases involving fluid structure interactions are investigated using several examples. A partitioned procedure involves several operators ensuring code coupling. A special attention must be paid to energy conservation at the fluid-structure interface, especially when it is moving and when strong non-linear behaviour occurs in both fluid and structure systems. In the present work, several fluid-structure code-coupling schemes are compared and discussed in terms of stability and energy conservation properties. The criteria are based on the evaluation of the energy that is numerically created at the fluid-structure interface. This is achieved by considering the staggering process due to the time lag between the fluid and structure solvers. Comparisons are made, and finally the article gives recommendations for creating a tool devoted to coupled simulations of fluid structure interactions. r (E. Longatte).
Fluid–structure interaction for aeroelastic applications
Progress in Aerospace Sciences, 2004
The interaction between a flexible structure and the surrounding fluid gives rise to a variety of phenomena with applications in many areas, such as, stability analysis of airplane wings, turbomachinery design, design of bridges, and the flow of blood through arteries. Studying these phenomena requires modeling of both fluid and structure. Many approaches in computational aeroelasticity seek to synthesize independent computational approaches for the aerodynamic and the structural dynamic subsystems. This strategy is known to be fraught with complications associated with the interaction between the two simulation modules. The task is to choosing the appropriate models for fluid and structure based on the application, and to develop an efficient interface to couple the two models. In the present article, we review the recent advancements in the field of fluid-structure interaction, with specific attention to aeroelastic applications. One of the key aspects to developing a robust coupled aeroelastic model is the presence of an efficient moving grid technique to account for structural deformations. Several such techniques are reviewed in this paper. Also, the time scales associated with fluid-structure interaction problems can be very different; hence, appropriate time stepping strategies and/or sub-cycling procedures within the individual field need to be devised. The flutter predictions performed on an AGARD 445.6 wing at different Mach numbers are selected to highlight the stateof-the-art computational and modeling issues. r
An Algorithm for the Strong-Coupling of the Fluid-Structure Interaction Using a Staggered Approach
We present a staggered approach for the solution of the piston fluid-structure problem in a timedependent domain. The one-dimensional fluid flow is modelled using the nonlinear Euler equations. We investigate the time marching fluid-structure interaction and integrate the fluid and structure equations alternately using separate solvers. The Euler equations are written in moving mesh coordinates using the arbitrary Lagrangian-Eulerian ALE approach and discretised in space using the finite element method while the structure is integrated in time using an implicit finite difference Newmark-Wilson scheme. The influence of the time lag is studied by comparing two different structural predictors.
Stable and accurate loosely-coupled scheme for unsteady fluid-structure interaction
AIAA Paper, 2007
This paper presents a new loosely-coupled partitioned procedure for modeling fluid-structure interaction. The procedure relies on a higher-order Combined Interface Boundary Condition (CIBC) treatment for improved accuracy and stability of fluid-structure coupling. Traditionally, continuity of velocity and momentum flux along interfaces are satisfied through algebraic interface conditions applied in a sequential fashion, which is often referred to staggered computation. In existing staggered procedures, the interface conditions undermine stability and accuracy of coupled fluid-structure simulations. By utilizing the CIBC technique on the velocity and momentum flux boundary conditions, a staggered coupling procedure can be constructed with similar order of accuracy and stability of standalone computations. Introduced correction terms for velocity and momentum flux transfer can be explicitly added to the standard staggered time-stepping stencils so that the discretization is well-defined across the deformable interface. The new formulation involves a coupling parameter, which has an interval of well-performing values for both classical 1D closed-and open-elastic piston problems. The technique is also demonstrated in 2D in conjunction with the common refinement method for subsonic flow over a thin-shell structure.
23rd AIAA Computational Fluid Dynamics Conference, 2017
Fluid structure interaction (FSI) describes a problem when a solid structure deforms or oscillates by the influence of the fluid flow, and thus a two-way interaction occur, such as in wind turbines, airfoils, parachutes, biological systems, including aneurysms, etc. One of the major challenges in the numerical simulation of this problem is the computational cost, and most of the current solvers are using an implicit method for fluid, with Newton-Raphson method being the most popular. However, an explicit method is relatively cheap. In this paper, explicit method with sub-iterations is compared with a Newton-Raphson method through an FSI benchmark case. We concentrate on comparison of accuracy as well as the computational performance of the two methods.
A hybrid finite-volume-rom approach to non-linear aerospace fluid-structure interaction modelling.
A fully-coupled partitioned fluid-structure interaction (FSI) scheme is developed for sub-and transonic aeroelastic structures undergoing non-linear displacements. The Euler equations, written in an Arbitrary Lagrangian Eulerian (ALE) coordinate frame, describe the fluid domain while the structure is represented by a quadratic modal reduced order model (ROM). A Runge-Kutta dual-timestepping method is employed for the fluid solver, and three upwind schemes are considered viz. AUSM +-up, HLLC and Roe schemes. The HLLC implementation is found to offer the superior balance between efficiency and robustness. The developed FSI technology is applied to modelling non-linear flutter, and the quadratic ROM demonstrated to offer dramatic improvements in accuracy over the more conventional linear method.
MODELING FLUID STRUCTURE INTERACTION FOR AEROSPACE APPLICATIONS
An approach for solving Fluid Structure Interaction in aerospace application is presented in this paper. The proposed approach is based on the two-way coupling between CFD code FlowVision and FEA code ABAQUS. The codes are coupled directly without using any 3 rd party software or intermediate structure. A direct link offers a full control over the load transfer and interpolation error free data exchange between the codes. The direct link is implemented using special meshing techniques (submerged meshes in FlowVision). FE mesh is subtracted from the Cartesian CFD mesh; all links between CFD mesh cell and outside faces of the finite elements are preserved. Node displacements are transferred directly between FlowVision and Abaqus without any interpolations. The above approach is illustrated with simulation of helicopter emergency landing on the water surface (helicopter equipped with flexible landing ballonets). The simulation objective is to estimate maximum loads on the helicopter hull caused by splashdown. The ballonets should absorb some of the impact and decrease acceleration on the helicopter crew. Results of two simulations are compared: helicopter lands on a rigid surface (ground) and on the still water surface (splashdown).
A methodology and computational system for the simulation of fluid-structure interaction problem
Journal of the Brazilian Society of Mechanical Sciences and Engineering, 2005
In this paper a flexible finite element computational tool developed to investigate fluidstructure interaction applications in two dimensions is described. We consider problems which can be modelled as a viscous incompressible fluid flow and a rigid body-spring system interacting nonlinearly with each other. The coupling is dealt with the use of an interface approach, in which each physics involved is solved with different schemes and the required information is transferred through the interface of both systems. This approach is, at least in principle, very flexible and computationally efficient as the best available scheme can be adopted to solve each physics. Here, a stabilized FEM considering the "ALE" (Arbitrary Lagrangian-Eulerian) formulation with Crank-Nicholson timeintegration is employed for the fluid-dynamics analysis, and the Newmark Method is used for the structural dynamics. Several important tools were incorporated into our system including different possibilities for the mesh movement algorithm, the computational domain decomposition into regions with and without mesh deformation, and the remeshing strategy (either global or local) to keep the necessary mesh quality. As application we present a study of the forced lock-in phenomena and a preliminary investigation on the suppression (or at least the reduction) of the vortex induced vibrations (VIV) on a solid circular cylinder using an idealization of a periodic acoustic excitation.
Analysis of some partitioned algorithms for fluid-structure interaction
Engineering Computations, 2010
Purpose-The purpose of this paper is to analyse algorithms for fluid-structure interaction (FSI) from a purely algorithmic point of view. Design/methodology/approach-First of all a 1D model problem is selected, for which both the fluid and structural behavior are represented through a minimum number of parameters. Different coupling algorithm and time integration schemes are then applied to the simplified model problem and their properties are discussed depending on the values assumed by the parameters. Both exact and approximate time integration schemes are considered in the same framework so to allow an assessment of the different sources of error. Findings-The properties of staggered coupling schemes are confirmed. An insight on the convergence behavior of iterative coupling schemes is provided. A technique to improve such convergence is then discussed. Research limitations/implications-All the results are proved for a given family of time integration schemes. The technique proposed can be applied to other families of time integration techniques, but some of the analytical results need to be reworked under this assumption. Practical implications-The problems that are commonly encountered in FSI can be justified by simple arguments. It can also be shown that the limit at which trivial iterative schemes experience convergence difficulties is very close to that at which staggered schemes become unstable. Originality/value-All the results shown are based on simple mathematics. The problems are presented so to be independent of the particular choice for the solution of the fluid flow.