1

Benchmarkon the numerical simulation

of a tube bundle vibration under cross flowFabien Huvelin (University of Lille & EDF R&D)

Marcus Vinicius Girao de Morais (University of Cergy-Pontoise & CEA)Franck Baj (CEA)Jean-Paul Magnaud (CEA)Elisabeth Longatte (EDF R&D )M’hamed Souli (University of Lille)

2

INTRODUCTION

CREATIF Program :Comprehension of the vibratory Response of an Array of Tubes in Interaction

with a Fluid.

water

steam

3

Experiment AMOVI (CEA)

L=70 mm

H=100 mm

D=12.15 mm

Strain gage

• 24 fixed tubes (Plexiglas, f>300Hz)• 1 moving tube (steel, f=14.3Hz)• P/D = 1.44• Single phase flow (flow rate from 0 to 5 m3.h-1) contact : fbaj@cea.fr

4

CONTENTS

1 – Methodology

2 – Real case (2D model)

3 – Numerical case (2D model)

4 – Real case (3D model)

5

1Methodology

6

1.1 Boundary conditionsMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

Boundary conditions at fluid-structure interface

⋅==

⋅ nu

nu

f

f

s

s

σσ

Partitioned coupling

Coupling scheme

LoadDisplacement/Velocity

FLUIDSOLVER

STRUCTURESOLVER

7

1.2 ToolsMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

Structure solver (rigid body)

• 1 degree of freedom oscillator for EDF tool• Finite element method with Lagrangian formulation for CEA tool

Fluid solver

• Finite volume method with Eulerian formulation (EDF tool)• Finite element method with Eulerian formulation (CEA tool)⇒ Use of an ALE technique to follow the structure interface (both cases)

Partitioned procedure

• Fluid solver needs structure data at time tn+1

• Structure solver needs fluid data at time tn+1

⇒ Prediction : the position of the structure interface at time tn+1

8

1.3 Improved serial staggered procedureMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

Fluid solverMesh updating

Structure solver

Prediction of the fluid structure interface

next

tim

e st

ep

Initialization

nstructure

nstructure

npredfluid VtXX

21, ∆+=+

Prediction of the fluid forceacting on the structure

npredstructurefluid

npredstructure FFF ,1, 2 −=+

(Piperno et al. ,2001)

9

2Real case (2D model)

10

2.1 Geometry (experiment AMOVI, CEA)IN

LET

FLO

W

WALL

WALLMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

FREE

OU

TLET

MOVINGTUBE

11

2.1 Geometry (experiment AMOVI, CEA)METHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

WALL

INLE

T FL

OW

WALL

FREE

OU

TLET

MOVINGTUBE

12

2.2 ParametersMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

12.15(mm)Diameter

0.298(kg.m-1)Mass

0.25(%)Natural damping

14.3(Hz)Natural frequency

Structure properties

10-3(kg.m-1.s-1)Dynamic viscosity

103(kg.m-3)Density

Fluid properties

[0.03 ; 0.15](m.s-1)Inlet velocity1.44(-)P/D

[1200;6000](-)Reynolds number[0.5 ; 2.8](-)Reduced velocity

13

2.3 Quiescent fluid : water frequencyMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

11.8

Analytical

11.55

EDF

11.63

CEA

(Hz)Water frequency

14

2.4 Quiescent fluid : water dampingMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

1.18

Analytical

1.215

EDF

1.216

CEA

(%)Water damping

extrapolation

15

2.5 Vibration under cross-flowMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

Re = 1200 Re = 3000 Re = 4800

• Numerical problem ?

• Physical problem ?

16

3Numerical case (2D model)

17

3.1 ParametersMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

10.00(mm)Diameter0.298(kg)Mass

0.04372.5

(%)(Hz)

Natural dampingNatural frequency

Structure properties

10-3(kg.m-1.s-1)Dynamic viscosity

103(kg.m-3)Density

Fluid properties

[0.001 ; 0.035](m.s-1)Inlet velocity1.44(-)P/D

[30 ; 1200](-)Reynolds number[0.1 ; 4.5](-)Reduced velocity

18

3.2 DisplacementMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

)( 13.0m.s 001.0 -1

−==

red

inlet

vv

)( 62.2m.s 02.0 -1

−==

red

inlet

vv

)( 93.3m.s 03.0 -1

−==

red

inlet

vv

19

3.3 Frequency and dampingMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

20

4Real case (3D model)

21

4.1 GeometryMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

• 7.5 Millions of elements • 2048 processors

22

4.2 Preliminary resultsMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

Displacement with 3D simulation

Displacement with 2D simulation

Reynolds = 3000 and DNS

23

4.3 3D effectMETHODOLOGY | REAL CASE (2D MODEL) | NUMERICAL CASE (2D MODEL) | REAL CASE (3D MODEL)

3D effects

Pressure Velocity

24

CONCLUSION

Catch the 3 behaviors of the structure

Used of an extrapolation method for the damping

Over-estimation of force fluid with 2D simulation (3D simulation is required)

3D simulation for numerical case

3D simulation with turbulent model for real case

Comparison of the real case to the experiment AMOVI (CEA)