CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   fluid mass in coupled FSI system (

Dr. V. Kumar July 11, 2007 09:19

fluid mass in coupled FSI system

Does anybody know if in a FSI simulation, in addition to forces exerted by the fluid, mass of the fluid domain is transferred to the FEM solver. There is no information about it in the ANSYS manuals. As far as I found out that only fluid forces are transferred across the interface.

Can anybody help in the matter?

thanks in advance

Peter Attar July 11, 2007 09:33

Re: fluid mass in coupled FSI system
From a strictly theoretical standpoint I don't see why this would be necessary for a truely two-way coupled FSI problem. The force is transferred in order to satisfy force equilibrium at the interface. The displacements are transferred to satisfy kinematic boundary conditions. What conditions are you trying to satisfy by transferring fluid mass?

Gert-Jan July 11, 2007 13:25

Re: fluid mass in coupled FSI system
Do you mean the density of the fluid and its damping character?


V. Kumar July 11, 2007 13:56

Re: fluid mass in coupled FSI system
yes indeed, the density would influence the dynamical characteristics of the structure. The combined mass of the system increases due to the presence of fluid which e.g. will change the eigen-frequency of the system.

Peter Attar July 11, 2007 15:19

Re: fluid mass in coupled FSI system
The only reason you would need to do this would be if you were not actually simulating the fluid. With two way FSI this effect should be taken care of by directly modeling the fluid and structural equations...try it. Blevins has some theoretical/experimental computations for fluid loading plates. Put your plate in a fluid and give it an initial motion. An FFT of the response should show that the response frequencies are changed due to the coupling with the fluid.

V. Kumar July 12, 2007 05:31

Re: fluid mass in coupled FSI system
Do you have a reference where it is shown that the structual equations take fluid density effect into account?

Frequency response could also be changed due to fluid forces (e.g. pressure or viscous forces) exerted on the plate. This is not a real justification for the density effects.

Peter Attar July 12, 2007 08:53

Re: fluid mass in coupled FSI system
You seem to be missing my point or intentionally ignoring it. What I am saying is that from a continuum mechanics standpoint the equations which govern the deformation of the structural domain do not include the density of the fluid. Period. Any papers which deal with fluid-structure interaction will show this. From a pure theory of elasticity standpoint, your fluid-structure interaction problem consists of Cauchy equations whose constitutive equations are appropriate for the structure, Cauchy equations with constitutive equations appropriate for the fluid and the conditions which must be satisfied at the boundary between the two domains, namely stress and kinematic boundary conditions.

Where are you getting the idea that in a true fluid-structure interaction problem the structural equations of motion need to somehow include the fluid density?

V. Kumar July 12, 2007 10:17

Re: fluid mass in coupled FSI system
Forget all equations. I am not talking about a general FSI algorithm.

Think of an oscillating pipe containing fluid, now depending on which fluid it contains (or passes through it), eigen modes of the pipe will change. This is a pure static effect, i mean it is not a fluid dynamical effect. This is the effect I am talking about. The kinematic and surface forces (stress) boundary conditions has nothing to do with this. My only concern is that do I have to include this effect manually or somehow ANSYS is taking care of this effect.

Peter Attar July 12, 2007 10:35

Re: fluid mass in coupled FSI system
ANSYS somehow takes care of it...BY SOLVING THE CORRECT EQUATIONS!!!! Think of it this way. If you had a pipe with some foam material attached to the inside would you need to add the density of the foam to the density of the pipe? No. The only thing you would need to do would be to correctly identify the Lame coefficients for the foam in order to model correctly the overall modal properties. Right? Having a fluid sitting in there is no different.

Are you trying to run a modal analysis with the fully coupled FSI system? I don't think you can do this. Since the formulation in ANSYS is not monolithic. In this case YOU WOULD have to artificially add the fluid mass. However you made no mention of modal analysis so my analysis has been based on the assumption that you wanted to do a coupled transient FSI simulation.

V. Kumar July 12, 2007 12:10

Re: fluid mass in coupled FSI system
thanks for the reply, however, I am still not conviced about what ANSYS does. I will run a case myself to check this and may be post here.

Yes I know you can not run FSI with ASNYS modal analysis since ANSYS's modal algorithm solves equation in freuency domain and CFX in time-domian. I was not trying for direct Modal analysis. During a transient FSI run if density of the fluid changes, one would like to see such effects.

However, you always can do eigen-mode analysis with coupled FSI by running a transient simulation and taking a fourier transform of the transient signal to obtain the eigen frequency.

Robin July 16, 2007 10:23

Re: fluid mass in coupled FSI system
Hi Kumar,

Peter is right. When ANSYS solves an FSI analysis it is solving the structural system of equations using the Finite Element Method (ANSYS) and the fluid dynamics equations by a Element Based Finite Volume Method (CFX). Whether fluid or solid, each solver is accounting for the mass, momentum and elastic effects of the materials. A conservative mapping of forces and displacements is done at the interface during stagger iterations performed several times within a timestep to ensure the system of equations is solved implicitely.

The forces returned by CFX are a result of the fluid momentum and the mass response of the system. If the material created or used in the CFX portion of the analysis includes compressibility (which can be included for both liquids and gases), these effects will be represented in the solution, assuming appropriate care is taken to reduce spatial and temporal discretization errors (i.e. grid and time step must be chosen appropriately).

Regards, Robin

All times are GMT -4. The time now is 15:36.