Computational Fluid-Structure

Interaction
AE4117

Sander van Zuijlen

Assistant Professor
Faculty of Aerospace Engineering
Aerodynamics Group, Building 64, room 0.39

Fluid-Structure Interaction (FSI)

• Structure of the course: 6 or 7 lectures:

13-2 Introduction, Fluid-mechanics (LN – Ch. 1 & 2)

20-2 Spatial coupling, mesh motion (LN – Ch. 4 & 5)

27-2 Time dependent problems, ALE, D-GCL (LN – Ch. 3, HO – Ch. 1&2)
6-3 Time dependent problems, ALE, D-GCL (LN – Ch. 3, HO – Ch. 1&2)
13-3 Partitioned temporal coupling (HO – Ch. 3)

20-3 Strong coupling, sub-iterations

27-3 TBD

LN : Lecture Notes LectureNotes_ps2pdf_6Mai.pdf

HO: Notes on high order time integration Notes_HighOrderTimeIntegration.pdf

1
Fluid-Structure Interaction (FSI)
• Examination: two exercises and oral examination
Exercises: (40% of grade)
First exercise available on Brightspace 20-2 (deadline hand-in 13-3)
Second exercise available on Brightspace 6-3 (deadline hand-in 27-3)
Exercises can be done in groups of maximum 3 students.
Missing the deadline:
One week late: 0,5 point deduction of exercise grade

> one week late: 1,0 point deduction of exercise grade

Oral exam (30-45 mins): (60% of grade)

You can only do the oral examination after handing in the assignments
Prepare small presentation (15 mins) on an FSI related topic of your choice
Oral examinations will be scheduled period 4-4 to 18-4 and 7-5 to 10-5
period 17-6 to 28-6
To pass the course you need 6,0 or higher for the oral exam part!

Fluid-Structure Interaction (FSI)

• Today:
• Examples of real-world and computational FSI

• Introduction of the coupled problem

• Concepts of Lagrangian and Eulerian reference frames

• Conservation laws for fluids

• Basics of Finite Volume discretization

2
Examples of possibly dangerous FSIs

Fluid-structure interaction also relevant

for wind turbines

3
FSI: (Dynamic) interaction between flows
and deforming structures

• How to model this numerically?

• Which physics do we need to model?

• Where does the interaction between flow and structure occur?

• Which conditions should be satisfied?

FSI: (Dynamic) interaction between flows

and deforming structures

• In all examples there is a dynamic, possibly dangerous, interaction

between the flow around a deforming structure

• Multi-physics are involved (solid mechanics & fluid mechanics)

• Aerodynamic loads on the structure cause a deformation of the structure
• Deformation of the aerodynamic surface results in a change in
aerodynamic loads

• Coupling between flow and structure at the fluid-structure interface:

• Equality of velocity at (and location of) the interface
• Equilibrium in stress on the interface

4
Simulation of FSI of aircraft

Another FSI simulation

5
Vortex shedding behind flapping wing

FSI from vortex shedding behind cylinder

6
Steady and flutter behavior of a plate

FSI: requires solving the fluid on a

deforming domain

14

7
FSI: requires solving the fluid on a
deforming domain

15

FSI: solving a coupled system

CFD Flow computation

IN: Shape of the wing

FSI pressure forces/loads

OUT:

CSM forces coupling deformation

Structure computation
CSM IN: forces on wing

OUT: deformation of wing

16

8
Coupling diagram of flow and structure

Interface Moving
displacement domain

Structure Flow mesh Flow

Interface forces

Coupling diagram of flow and structure

How to interpolate
between meshes
Interface Moving How to solve
displacement How to deform
fluid mesh
domain fluid flow
equations

Structure Flow mesh Flow

How to couple in … on a
time deforming mesh

Interface forces

How to interpolate
between meshes

9
 Coupled fluid and structure equations

• Structural system:
∂2 q ∂q
M 2
+ D + Kq = Finterface
∂t ∂t

 
d dx 
• Fluid system: ∫ W dV + 
∫ S(t ) (F(W ) − W dt ) ⋅ n dS = 0
dt V (t )

• Moving fluid mesh

• Boundary conditions at the interface:

• flow speed = time derivative of displacement of the structure

• stress in structure at interface = pressure of flow at interface

Computation of unsteady flows on fixed

meshes

• Conservation laws:
• Conservation of mass
• Conservation of momentum
• Conservation of energy

• Navier-Stokes equations

• Finite volume discretisation

10
 Conservation principles, control mass

• A conservation law relates the rate of change of an extensive property

(such as mass, or momentum) in a given control mass (a parcel of
matter) to externally determined effects

• Mass is neither created nor destroyed, so

Dm
|CM = 0
Dt

• Momentum can be changed by the action of forces, so

! !
Dm v
|CM = å F
Dt

Conservation principles, control volume

• In fluid flows it is more convenient to deal with a control volume,
rather than with a control mass, which quickly passes through the
region of interest

• If φ is a conserved intensive property, then the corresponding

extensive property is:
F= ò r j dW
W CM

• Conservation equation for a constant control volume (CV) can be

written as:
Dmφ D d  
Dt
|CM = ∫ ρ φ dΩ = dt ∫ ρ φ dΩ + ∫ ρ φ v ⋅ n dS
Dt ΩCM ΩCV SCV

With ΩCV is the CV, SCV is the surface enclosing the CV, n is the unit normal orthogonal to S directed
outwards, v is the fluid velocity

11
Lagrangian versus Eulerian approach
• Lagrangian approach very suitable for problems with changing
geometry (mesh naturally follows the material)
• Lagrangian approach widely used in computational structure dynamics

• Eulerian approach very suitable for fixed geometry problems through

which a medium is flowing
• Eulerian approach widely used in computational fluid dynamics

• For fluid-structure interaction flow domain not constant geometry

anymore => Arbitrary Lagrangian Eulerian (ALE) approach

Conservation of mass on fixed meshes

• The integral form of the mass conservation (or continuity) equation,

follows from the general form by φ = 1:
d  
∫ ρ dΩ + ∫ ρ v ⋅ n dS = 0
dt ΩCV SCV

• Applying Gauss’ theorem to the convection term we can transform the

surface integral into a volume integral

• Allowing the CV to become infinitesimally small leads to the

differential form:
¶r !
+ Ñ × (r v ) = 0
¶t

12
Conservation of momentum on fixed meshes
• The integral form of the momentum
! conservation equation, follows
from the general form by φ = v :
d    
∫ ρ v dΩ + ∫ ρ v v ⋅ n dS = ∑ F
dt ΩCV SCV

• Forces which may act on the fluid in a CV:

– Surface forces T (pressure, normal and shear stress, surface tension, etc.)
– Body forces f (gravity, centrifugal and Coriolis forces, etc.), so
! ! !
å F = ò T × n dS + ò r f dW
S W

• For Newtonian fluids, the stress tensor T can be written as:

T=- p+τ
with τ the viscous part of stress tensor:
∂vi ∂v j 2 
τ ij = µ ( + ) − µδij ∇ ⋅ v.
∂x j ∂xi 3

Conservation of momentum on fixed meshes

• The corresponding equation for the i-th Cartesian component is:

d   
∫ ρ vi dΩ + ∫ ρ vi v ⋅ n dS = ∫ ( - p + τ i )⋅ n dS + ∫ ρ fi dΩ
dt ΩCV SCV SCV ΩCV

• In differential form this is:

∂ρ vi  ∂p
+ ∇ ⋅ ( ρ vi v) = − + ∇ ⋅ τ i + ρ fi
∂t ∂xi

Conservation of energy along similar lines

13
Resulting system of equations

• For viscous flows the Navier-Stokes equations in 3D:

– Conservation of mass
– Conservation of momentum (three)
– Conservation of energy
– Equation of state: p = f(ρ , E) (conserved variables)

• For turbulent flows: problem to resolve small scales, solutions:

Reynolds Averaged Navier-Stokes (RANS), Large Eddy
Simulation (LES), Direct Numerical Simulation (DNS)

• For inviscid flows the Euler equations are obtained, setting the
viscous stresses equal to zero

The Navier-Stokes equations on a fixed

mesh in conserved variables in 2D
 ! ! ! ! !
d       ¶W ¶E (W ) ¶F (W ) !
∫ W dΩ + ∫ [ E(W ), F(W )]⋅ n dS = ∫ S dΩ + + =S
dt ΩCV SCV ΩCV ¶t ¶x ¶y

érù é ru ù
ê ú
! ê ru ú r + - t
2
! ! u p
E (W ) = ê ú
xx
W =ê ú ê r u v - t xy ú
ê rv ú
ê ú ê ú
ëê ( rE + p ) u - u t - v t + q ú
ë rE û xx xy xû

é 0 ù é rv ù
! ê rf x ú ! ! ê r u v - t ú
S =ê ú F (W ) = ê
xy ú
ê rf y ú ê r v 2 + p - t yy ú
ê ! !ú ê ú
ë rf × v û ëê ( rE + p ) v - ut xy - vt yy + q ú
yû

14
Finite volume discretisation on fixed
meshes
• The finite volume methods uses the integral form of the conservation
equations as starting point

• The solution domain is subdivided into a finite number of Control

Volumes, and the conservation eqs, are applied to each CV

• At the centre of each CV lies a computational node at which the

variables are to be calculated.

• Interpolation is used to express variable values at the CV surface in

terms of the nodal (CV-center) values

• Surface and volume integrals are approximated using suitable

quadrature formulae

The FV approximation of surface integrals

• The net flux through the CV boundary is the sum of integrals over the four
(in 2D) CV faces:

ò f dS = å ò f dS ,
S k Sk
with, for example f = ( r j v × n)
• The simplest approximation is the midpoint rule which leads to second order:

ò f dS = f S
Se
e e » f e Se

• Another second order approximation can be obtained with the trapezoid rule:
Se
ò f dS » 2 ( f
Se
ne + f se )

• A fourth-order approximation is Simpson’s rule:

Se
ò f dS » 6 ( f
Se
ne + 4 f e + f se )

15
The FV approximation of volume integrals

• The simplest second-order accurate approximation is to replace the

volume integral by the product of the mean value and the control
volume:

ò q dW = å ò q dW = q ΔΩ » q
W k Wk
P ΔΩ

• An approximation of higher order requires the values at more

locations than just the center. These values need to be obtained by
interpolation.

Vortex shedding behind airfoil at high

angle of attack

16
Fluid-Structure Interaction (FSI)

• Today:
• Examples of real-world and computational FSI
• Introduction of the coupled problem

• Concepts of Lagrangian and Eulerian reference frames

• Conservation laws for fluids
• Basics of Finite Volume discretization

• Next week:
• Spatial coupling between meshes
• Mesh deformation

17