Identifying the frequency selection of fluid/structure instabilities - - PowerPoint PPT Presentation

identifying the frequency selection of fluid structure
SMART_READER_LITE
LIVE PREVIEW

Identifying the frequency selection of fluid/structure instabilities - - PowerPoint PPT Presentation

Identifying the frequency selection of fluid/structure instabilities when the interaction is large Olivier Marquet 1 & Lutz Lesshafft 2 1 Department of Fundamental and Experimental Aerodynamics 2 Laboratoire dHydrodynamique, CNRS-Ecole


slide-1
SLIDE 1

Identifying the frequency selection

  • f fluid/structure instabilities

when the interaction is large

Olivier Marquet1 & Lutz Lesshafft2

1 Department of Fundamental and Experimental Aerodynamics 2 Laboratoire d’Hydrodynamique, CNRS-Ecole Polytechnique

11th Euromech Fluid Mechanics Conference 12-16 September 2016, Sevillel, Spain

slide-2
SLIDE 2

Context

2

Flow-induced structural vibrations

Civil engineering Aeronautics Offshore-marine industry

Predict the onset of vibrations (based on stability analysis) and control them The origin of fluid/structure instabilities ?

slide-3
SLIDE 3

A model problem spring-mounted cylinder flow

3

=

  • = 40

Reynolds number

= ⁄

Structural frequency

  • Structural damping

=

  • Density ratio

One spring in the cross-stream direction

slide-4
SLIDE 4

Self-sustained oscillations

4

= 40 , = 10 , = 0

No oscillation

= 0.6 = 0.7

Weak oscillation

= 0.9

Strong oscillation

= 1.1

No oscillation

slide-5
SLIDE 5

Outlines

5

Titre présentation

1 – Stability analysis of the fluid/structure problem

a – Operator definition and formalism b – Results for weak and strong interaction

2 – Identification the driving dynamics of coupled modes

a – Operator decomposition approach b – Results for strong interaction

slide-6
SLIDE 6

Stability analysis of the coupled fluid/solid problem

6

()

  • (, )
  • = (! + #$) %
  • &
  • ′((, )) = ( , ) ( *+, - . + /. /.

Fluid/solid components Growth rate/frequency

Cossu & Morino (JFS, 2000)

Stability of the steady solution (fixed cylinder)

Damped harmonic oscillator Linearized fluid equations

slide-7
SLIDE 7

Results – Eigenvalue spectrum

7

= 0.75

= 101

Are the coupled modes driven by the fluid or the solid dynamics ?

Eigenvalue spectrum

slide-8
SLIDE 8

Results – Components of eigenvector

8

Fluid component Solid component

2 = 3 = 0 2 = 3 ≠ 0

Fluid (wake) mode Solid mode For small mass ratio - strong interaction ?

2 = 3 ≠ 0

Infinite mass ratio - weak interaction

slide-9
SLIDE 9

Results – Variation of stiffness

9

= 0.75 0.4 < < 1.2

Infinite mass ratio – weak interaction Solid mode : $ ∼ (= structural frequency) Wake modes : $ ∼ $8 (= vortex-shedding frequency)

slide-10
SLIDE 10

Results – Variation of stiffness

10

= 200 = 10

Finite mass ratio – strong interaction

Destabilization of solid branch Destabilization of fluid branch? The two branches exchange their « nature » for small mass ratio

Zhang et al (JFM 2015), Meliga & Chomaz (JFM 2011)

slide-11
SLIDE 11

Outlines

11

1 – Stability analysis of the fluid/structure problem

a – Operator definition and formalism b – Results (various mass ratio / structural frequency)

2 – Identification of the driving dynamics

a – Operator decomposition approach b – Results

slide-12
SLIDE 12

The eigenvalue problem

12

()

  • (, )
  • = (! + #$) %
  • &
  • Eigenvalue problem - Coupled operator

Infinite mass ratio

()

  • (, )
  • = (! + #$) %
  • &
  • Fluid mode = eigenvalue/vector of

Solid mode = eigenvalue/vector of

slide-13
SLIDE 13

From operator to eigenvalue decomposition

13

= 9 + : = ;

Operator decomposition

In general, is not an eigenmode of 9 or : , so

9 = ;9 + <

9

: = ;: + <

: Eigenvalue decomposition with residuals

<

9 ≠ 0, < : ≠ 0 but < 9 = −< : = <

;9 + ;: = ;

How to compute the eigenvalue contributions ;9/;: ?

slide-14
SLIDE 14

Computing eigenvalue contributions

14

Expansion of the residual on the set of other eigenmodes ? Orthogonal projection on the mode using the adjoint mode +

+@(9 ) = ;9( +@ ) + A <

? ( +@ ?) ?

= 1

Bi-orthogonality

= 0

Normalisation

9 = ;9 + A <

? ? ?

< = A <

? ? ?

;9 = +@(9 ) ;: = +@(: )

Adjoint mode-based decomposition

; = ;9 + ;:

slide-15
SLIDE 15

Why this particular eigenvalue decomposition?

15

For an identical decomposition of the operator,

  • ther eigenvalue decompositions are possible

; B9/: = ;9/: ± A <

? ( @ ?) ?

≠ 0

Non-orthogonal projection on the mode

; B9 = @(9 ) ; B: = @(: )

Direct mode-based decomposition

; = ; B9 + ; B:

9 = ;9 + A <

? ? ?

But it includes contributions from other eigenmodes

M.Juniper (private communication)

: = ;: − A <

? ? ?

slide-16
SLIDE 16

Application to the spring-mounted cylinder flow

16

; = ;

+ ;

Adjoint mode-based decomposition

;

= +@ ( + ) Fluid contribution

; =

+@( + ) Solid contribution

  • = ; %
  • &
  • Adjoint fluid component

Adjoint solid component

slide-17
SLIDE 17

Application to the spring-mounted cylinder flow

17

; = ;

+ ;

Adjoint mode-based decomposition

;

= ; ( +@ ) Fluid contribution

; = ;(

+@ ) Solid contribution

  • = ; %
  • &
  • Direct and adjoint

fluid components Direct and adjoint solid components

slide-18
SLIDE 18

Stability results for D = EFF – Solid branch

18

Growth rate Frequency

slide-19
SLIDE 19

Solid branch: Frequency decomposition

19

$ = $ + $

Fluid Solid

$ = ℑ(;)

Fluid contribution Solid contribution

$ = ℑ(;)

The frequency is selected by the solid dynamics

slide-20
SLIDE 20

Solid branch: Growth rate decomposition

20

! = ! + !

! = ℜ(;)

Fluid contribution Solid contribution

! = ℜ(;)

Large and opposite contributions in the unstable region

slide-21
SLIDE 21

Solid branch: Growth rate decomposition

21

! = ! + !

! = ℜ(;)

Fluid contribution Solid contribution

! = ℜ(;)

|!| > |!| destabilization by the solid contribution |!| > |!| destabilization by the fluid contribution

! > 0

Low stiffness High stiffness

slide-22
SLIDE 22

Local (fluid) contribution to the growth

22

Phase difference between and

  • +
  • Schmid & Brandt (AMR 2014)

Stabilizing region Destabilizing region

slide-23
SLIDE 23

Stability results for D = KF

23

SM WM

Frequency of the Wake Mode branch Eigenvalue spectrum

slide-24
SLIDE 24

« So called » fluid branch: Frequency decomposition

24

$ = $ + $

$ = ℑ(;)

Fluid contribution Solid contribution

$ = ℑ(;)

$ ∼ $

Frequency selection by the fluid dynamics

$ ∼ $

Frequency selection by the solid dynamics

$ ∼ $ Low stiffness High stiffness

slide-25
SLIDE 25

Conclusion

25

  • Operator decomposition approach applied to coupled fluid/solid

modes

  • No need to vary the parameters (mass ratio or stiffness), need to

determine the adjoint modes.

  • Results similar to « wavemaker » analysis (structural sensitivity)
  • Not a variation of eigenvalues but a decomposition of eigenvalues
  • Extension to more complex solid dynamics (Jean-Lou Pfister - PhD)

Thank you

slide-26
SLIDE 26

Pure modes (infinite mass ratio)

26

Titre présentation

;

  • = L
  • M
  • Fluid

modes

;∗ O O = L@

  • @

M@ O O

Direct Adjoint Solid modes

; = L = 0 ; = M (;& − L) = ;

  • ∗O

P = L@O P

;

  • ∗& − M@ O

P =

  • @O

P

;

∗O P = M@O P

O

P = 0

slide-27
SLIDE 27

Projection of coupled problem on pure fluid modes

27

Titre présentation

(; & − L) = (;& − M) = O

P@ ; & − L + O P@ ; & − M − O P@ = O P@

;∗O

P − L@O P @

+ ;∗O

P − M@O P −

  • @O

P @

= O

P@

; − ;

(O P@ ) + ; − ; (O P@ ) = O P@

; − ;

=

O

P@

(O

P@ + O P@ )

slide-28
SLIDE 28

Projection of coupled problem on pure solid modes

28

Titre présentation

(; & − L) = (;& − M) = O

P@ ; & − L + O P@ ; & − M − O P@ = O P@

;∗O

P − L@O P @

+ ;∗O

P − M@O P −

  • @O

P @

= O

P@

(; − ;)(O

P@ ) = O P@

; − ; = O

P@

O

P@

slide-29
SLIDE 29

Direct-based decomposition of the unstable mode

29

Titre présentation

Frequency Fluid Solid Growth rate

= 200

Solid Fluid

slide-30
SLIDE 30

Infinite mass ratio - Fluid Modes

30

; = ;

+ ;

Adjoint mode-based decomposition

;

= +@ ( + ) Fluid contribution

; =

+@( + ) Solid contribution

Fluid Modes

QR = F ;

= +@ = ;

  • F
  • = ; %
  • &
  • = 0

= 0 ; = 0 OK

slide-31
SLIDE 31

Infinite mass ratio - Structural Mode

31

; = ;

+ ;

Adjoint mode-based decomposition

;

= +@ ( + ) Fluid contribution

; =

+@( + ) Solid contribution

Structural Mode

;

= 0

  • F
  • = ; %
  • &
  • = 0

= 0 ; =

+@ = ;

OK QS

+ = F

slide-32
SLIDE 32

Infinite mass ratio – Direct mode-based decomposition

32

; = ; B

+ ;

B

Direct mode-based decomposition

; B

= QS T ( + ) Fluid contribution

; B = QR

T( + ) Solid contribution

Structural Mode

; B

= ; ( @ )

  • = ; %
  • &
  • = 0

= 0 ; B = ; (

@ )

NOT OK +

= ;

slide-33
SLIDE 33

Infinite mass ratio – Direct mode-based decomposition

33

; = ; B

+ ;

B

Direct mode-based decomposition

; B

= QS T ( + ) Fluid contribution

; B = QR

T( + ) Solid contribution

Structural Mode

; B

= ; ( @ )

  • = ; %
  • &
  • = 0

= 0 ; B = ; (

@ )

NOT OK

(large fluid response)

(;& − ) =

slide-34
SLIDE 34

Results for D = EFF − Solid Mode

34

SM WM

Growth rate (SM) Frequency (SM)

  • The frequency is quasi-equal to
  • The growth rate gets positive for close $8
slide-35
SLIDE 35

Wake Mode: growth rate decomposition

35

Growth rate Solid Fluid

= 10

Small destabilization due to the solid Large destabilization due to the fluid

slide-36
SLIDE 36

Wake Mode: growth rate decomposition

36

Growth rate Solid Fluid

= 10

Small destabilization due to the solid Large destabilization due to the fluid

slide-37
SLIDE 37

Wake Mode: growth rate decomposition

37

Growth rate Solid Fluid

= 10

Small destabilization due to the solid Large destabilization due to the fluid

slide-38
SLIDE 38

Free oscillation

= 40 ; = 50

No oscillation

$ = 0.60 $ = 0.66

Weak oscillation

$ = 0.90

Strong oscillation

$ = 1.10

No oscillation

slide-39
SLIDE 39

Solid displacement – Fluid fields

Multiple solutions

slide-40
SLIDE 40

Methods for identifying the dynamics of coupled modes

40

  • Energetic approach (Mittal et al, JFM 2016)

Transfer of energy from the fluid to the solid component

  • Classical wavemaker analysis (Giannetti & Luchini, JFM 2008, …)

Structural sensitivity analysis of the eigenvalue problem. Largest eigenvalue variation induced by operator perturbation

  • Operator/Eigenvalue decomposition