Active Control of Flow over a Circular CylinderTurbulence and Flow Control Seoul National University Center for Research Contents 1. Characteristics of flow over a circular cylinder

Turbulence and Flow Control Seoul National UniversityCenter for

Research

Active Control of Flow over a Circular Cylinder

Haecheon Choi and Jinsung Kim

Center for Turbulence and Flow Control ResearchSchool of Mechanical and Aerospace Engineering

Seoul National University

http://tfc.snu.ac.kr/

FLOWCOMAG, Dresden, April 1-2, 2004


Research

Contents

1. Characteristics of flow over a circular cylinder

2. Distributed forcing (active control)

3. Wake disrupter (passive control)

4. Conclusions


Research

Flow characteristics around a cylinder• Steady flow (Re<47)• 2-dimensional unsteady laminar flow

(47<Re<160)

• Three-dimensional laminar flow: Re=220 (mode A; )

• Three-dimensional laminar flow:Re=300 (Mode B; Re > 230)

• Turbulent flow in the wake:Re=3900

Numerical method (Choi, Moin & Kim, JFM 1993)

230160 ≤≤ Re


Research

Flow characteristics around a sphere

• Re=250(steady planar symmetric)

• Re=100(steady symmetric)

• Re=425(unsteady asymmetric)

• Re=300(unsteady planar symmetric)

• Re=3700(turbulent flow

in the wake)

• Re=104

(turbulent flow in the wake)

Immersed boundary method (Kim, Kim & Choi JCP 2001)


Research

Motivation

• Flow over a bluff body produces a significant amount of drag force and lift fluctuations and also generates flow-induced noise due to vortex shedding. Thus, there have been large amount of efforts devoted to control the flow around bluff bodies and to reduce the drag and lift.

• Many active and passive controls that are homogeneous in the spanwise direction have been presented: base bleeding, splitter plate, secondary small cylinder, rotation, single frequency forcing, electromagnetic forcing, some active blowing/suction based on sensing of flow variables (e.g. Min & Choi, JFM 1999).

Direct interaction with mean flowOr with two-dimensional instability


Research

Turbulent flow over a backward-facing step (Kang & Choi, JFM 2002): 80% mixing enhancement using suboptimal control

Blowing/suction profile Sine curveActive open-loop control

(Distributed forcing)

• Wall pressure sensing

• Blowing/suction at the tip

• Generation of strong three-dimensional flow

Motivation

Laminar flow over a backward-facing step (Choi, Hinze & Kunisch, ANM 1999)


Research

Objectives

• In the present study, the possibility of applying an active control, which is varying in the spanwise direction (i.e. distributed forcing), to flow over a bluff body for drag reduction is investigated for a wide range of Reynolds numbers.

• Based on a successful active control method, we also develop a passive devicefor reduction of drag on a bluff body.

Circular cylinderModel vehicle


Research

Distributed forcing : cylinder

• Boundary conditions- uniform flow at inlet and far-field- convective outflow- periodic in the spanwise direction- blowing/suction at top and bottom slots

• Grid system : C-type grid

• Schematic of forcing- forcing profile : steady in time & periodic in z

- forcing phase : in-phase forcing out-of-phase forcing

=

z

zzλ

πφφ 2sin)( max

zλ zλ

∞U

d


Research

Distributed forcing : cylinder• Effects of the spanwise wavelength ( ), forcing phase (in-phase vs. out-

of-phase) and forcing amplitude ( ) on the flow are considered for Re=100.

• Reynolds numbers consideredRe=40 : 2D, steady flowRe=80,100,140 : 2D, unsteady flowRe=220 : 3D unsteady flow with Mode-A instabilityRe=300 : 3D unsteady flow with Mode-B instabilityRe=3900 : turbulent wake (LES)

• Number of grid pointsRe < 140 : 320 x 120 (x 16)Re = 220, 300 : 320 x 120 x 48Re = 3900 : 672 x 160 x 64

zλmaxφ


Research

• Effect of the spanwise wavelength

• Other parameters :

100 200 300

1.1

1.15

1.2

1.25

1.3 1d2d3d4d5d6d10d

DC

time

∞= U.max 10φ , in-phase forcing

• Time history of the drag coefficient

- Drag is minimum at = 4 to 5.dz /λ

- Maximum drag reduction is 20 %

This wavelength is similar to that of the mode-A instability.

-

zλ

cf. Wavy cylinder by Darekar & Sherwin (2001, JFM): = 5.6 (16% DR).dz /λ

Distributed forcing : cylinder (Re=100)


Research

dz 1=λ

dz 2=λ

dz 3=λ

dz 4=λ

dz 5=λ

dz 6=λ

dz 10=λ

• Instantaneous flow structures (iso-surface of = - 0.02) 2λ• Parameters : ∞= U.max 10φ , in-phase forcing



Research

• Effect of the phase difference between upper and lower forcing• Other parameters :

0 10 20 30 40 50 601.20

1.22

1.24

1.26

1.28

1.30

1.32

1.34

out-of-phase forcingin-phase forcing


DC

time

in-phase forcingout-of-phase forcing

- In-phase forcing

- Out-of-phase forcing• Vortical structures

dU z 5,1.0max == ∞ λφ



Research

100 200 300 400

1.1

1.15

1.2

1.25

1.3

0 0.05 0.1 0.151.00

1.05

1.10

1.15

1.20

1.25

1.30

1.35

1.40

• Effect of the forcing amplitude• Other parameters : dz 5=λ , in-phase forcing

maxφ

• Time history of the drag coefficient • Mean drag coefficient

time

DC

∞= U01.0maxφ

05.0

07.0

08.0

10.0

15.0

DC

∞U/maxφ

saturation



Research

0 1 2 3 4 5-0.75

-0.70

-0.65

-0.60

-0.55

-0.50

-0.45

-0.40

-0.35

-0.30

-0.25

0 1 2 3 4 5106107108109110111112113114115116117118119120121122

• base pressure coefficient • separation point

• Spanwise variations of the base pressure coefficient and the separation point for the in-phase forcing

∞= U01.0maxφ

05.0

07.0

15.0,10.0,08.0

∞= U01.0maxφ

05.0

07.0

08.0

10.015.0

bpC sepθ

dz / dz /

25.1/ =dz

75.3/ =dz



Research

0 100 200 300 400

1

1.1

1.2

1.3

1.4

4d5d6d

0 100 200 300

1.16

1.18

1.20

1.22

1.24

1.26

1.28

1.30

1.32

1.34

1.36

4d5d6d

DC

time time

DC

• Time history of the drag coefficient, Re=80 • Time history of the drag coefficient, Re=140

- Drag is minimum at = 5 to 6dz /λ - Drag is minimum at = 4 to 5dz /λ

• The base flow is 2-D laminar with Karman vortex shedding • Parameters : ∞= U.max 10φ , in-phase forcing

Distributed forcing : cylinder (Re=80, 140)


Research

0 50 100 150 2001.50000

1.50025

1.50050

1.50075

1.50100

1.50125

1.50150

1.50175

1.50200

time

DC

• Time history of the drag coefficient, Re=40

• When Re = 40, there is no vortex shedding and distributed forcing does not work.

• Flow structures



Research

550 600 650 700 7501.00

1.05

1.10

1.15

1.20

1.25

1.30

1.35

1.40

1.45

1.50uncontrolledinphase, 3.75dinphase, 1.875d

• At Re=220 (mode A),

DC

• Time history of the drag coefficients • Flow structure Uncontrolled

In-phase,

In-phase,

time

d.z 753=λ

d.z 8751=λ

∞= U.max 10φ



Research

100 125 150 175 200 225

1.00

1.10

1.20

1.30

1.40

uncontrolledinphase, 1dinphase, 3dout-of-phase, 3d

• At Re=300 (mode B),

DC

• Time history of the drag coefficients • Flow structure Uncontrolled

In-phase,

time

dz 3=λ

∞= U.max 10φ



Research

130 140 150 160 170 180 190-1.00

-0.75

-0.50

-0.25

0.00

0.25

0.50

0.75

1.00

130 140 150 160 170 180 1900.70

0.75

0.80

0.85

0.90

0.95

1.00

1.05

1.10

1.15

1.20

1.25

1.30

• Turbulent flow


DC

time

• Time history of the lift coefficient

LC

time

In-phase forcing

Out-of-phase forcing

In-phase forcingOut-of-phase forcing

UncontrolledIn-phase forcingOut-of-phase forcing

Uncontrolled

∞= U.max 10φ3900=Re /z dλ π=,,• Parameters :



Research

Instantaneous flow structures

• uncontrolled • Out-of-phase forcing • In-phase forcing



Research

• Small zλ

195 200 205 210-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

100 105 110-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

90 95 100 105-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

t t t

vv v

B

S

B

SSB

∞U

• v trace at (x,y)=(2.0d, 0.6d)

• vortical structures (top view)

dz =λdz 2=λ

dz 3=λ

π19.0π068.0

(Re=100, , )∞= U1.0maxφ dz 3≤λVortex shedding frequency is the same at all the spanwise positions.Shedding process at maximum suction has a faster phase than that at maximum blowing.

negligible difference

Mechanism of drag reduction due to the in-phase forcing


Research

50 100-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

t=6

t=44

t=80

• Optimal zλ

v

t

B

S

B

S

B

S

π

blowingsuction

• v trace at (x,y)=(2.0d, 0.6d)

phase difference

(Re=100, , )∞= U1.0maxφ dz 5~4=λ

forcing starts, zero phase difference

Phase difference between the suction and blowing positions gradually approaches to .Vortex shedding is dislocated completely.

π



Research


• Parameters : , in-phase forcing , dz 5=λ

• Iso-surfaces of spanwise vorticity )8.0( ±=zω

∞= U.max 10φ


Research

150 200 250-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0 0.05 0.1 0.15 0.20

1E-07

2E-07

3E-07

4E-07

5E-07

• Large zλ

v

t St

E(v)

Stblowing

Stsuction

Stmodulation

• v trace at (x,y)=(2.0d, 0.6d) • Energy spectrum of v

blowingsuction

(Re=100, , )∞= U1.0maxφ dz 7=λ

Different shedding frequencies appear at different spanwise locations. - frequency at suction location is higher than that at blowing location- modulation occurs at Stmoulation = Stsuction – Stblowing



Research

• Out-of-phase forcing :- works only for thin separating shear layer- additional three-dimensionality in the

shear layer delays the vortex shedding - shedding processes are merely pushed downstream

• Contour of streamwise velocity in an y-z plane (x=0.5d)

in-phase forcing out-of-phase forcing

Re=100

Re=3900

• At Re=3900, shear layers become thinner and more susceptible to external disturbances

d

d

069.0/ ≈dδ277.0/ ≈dδ

for Re=3900

for Re=100(

Drag reduction due to the out-of-phase forcing


Research

• The objective is to examine the applicability of the distributed forcing to flow over a bluff body with a blunt trailing edge (fixed separation).

Flow over a model vehicle

(Tombazis & Bearman 1997)


Research

0//, =∂∂=∂∂= ∞ ywyvuu

Convectiveoutflow b.c.

4200/Re ==• ∞ νhu

x

y

z

15/3 ≤≤− hx29/29 ≤≤− hy4/0 ≤≤ hz

)(64)(240)(350 zyx ××

)36.1/( =hδ

Turbulentboundary

layer(Reθ =670)

h

0//, =∂∂=∂∂= ∞ ywyvuu

• Domain size:

• Number of grids:

∞u

∞u

Model vehicle

Turbulentboundary

layer(Reθ =670)

Computational details (LES)


Research

In-phase forcing


)2sin()( max zzzλπ

Φ=Φ

b=0.1h

°45

h xyΦ

Φ°45

zy

hz 4=λ

∞=Φ u1.0max

hz 4=λ

zy

∞=Φ u1.0max

Steady blowing/suction

Distributed forcing: model vehicle


Research

1 2 3

-0.25

-0.2

-0.15

0 4

-0.5

-0.4

-0.3

0%30%—Pressure recovery (%)

0.500.350.50

Out-of-phase forcingIn-phase forcingUncontrolled flow

: Uncontrolled flow

: In-phase forcing

: Out-of-phase forcing

-Cpb

z/h

Cpb (y=0)

blowing suction

Distributed forcing: model vehicle• Base pressure coefficient


Research

• Vortical Structures

In-phase forcing


Uncontrolled flow

Distributed forcing: model vehicle


Research

Freestream

h

W

8:1 ellipse rectangular

Trip

Experimental setup for distributed forcing

- W=300mm, h=60mm

- Reh=u∞h/ν =20000, 40000, 80000

- Trip wires

- Model vehicle

u∞

W=300mm

240mm 140mm

x

y

zh=60mm

u∞

(Tombazis & Bearman 1997)


Research

blowingsuction

fanblowing suction

Actuator for distributed forcing

• Two fans comprise 1 wavelength.

• Four fans are constructed within the entire span.

• The maximum velocity from the slit : 4.1m/s in the absence of freestream.

• Forcing wavelength λz≈ 2.5h.


Research

-0.7

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

0 0.2 0.4 0.6 0.8 1-0.7

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

0 0.2 0.4 0.6 0.8 1

Cpb

z/W

Base pressure along the spanwise direction

• Base pressure was clearly increased with control.• Base pressure was recovered more at the blowing location.• Spanwise variation of base pressure became milder at Reh=40000. • The percentages of base-pressure recovery are about 36% and 18% at Reh=20000 and 40000,

respectively.

Cpb

z/W

BlowingSuctionBlowingSuction

Reh=40000Reh=20000

Uncontrolled Controlled

BlowingSuctionBlowingSuction


Research

Wake disrupter – a new passive deviceDistributed forcing: three-dimensional disturbances to the flow with nominally

two-dimensional Kármán vortex shedding.

uncontrolled controlled

• Small-size rectangular body attached to a part of the trailing edge.

• Can be easily attached or removed.

• No control input is required.

Wake disrupter


Research

lylz

xz

y

Freestream

y

h

Wx

z

8:1 ellipse rectangular

Trip

Experimental setup for wake disrupter

- W=300mm, h=60mm

- Reh=U∞h/ν =20000, 40000, 80000

- Trip wires

- Model vehicle

U∞

- ly/h, lz/h = 0.017~0.66- 71 cases with different sizes of disrupters.- Spanwise average of the 18 pressure values along y=0.

wake disrupter


Research

Experimental setup for wake disrupter


Research

-4

0 8

2018

22

0 0.1 0.2 0.3 0.4 0.5 0.6 0.70

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Percentage of base-pressure increase (Reh=20000)Percentage of base-pressure increase = (Cp– Cpo)/Cpo ×100(Cpo : base-pressure coefficient

without disrupter)

lz / h

ly / h

Optimum size :

ly/h=0.2 ~ 0.25, lz/h=0.2 ~ 0.25

lz / θ403010 20

10

20

30

ly / θ

40

lz / δ1 3

1

2

2

ly / δ


Research

08

-2

14

2218

23

lz/h0 0.1 0.2 0.3 0.4 0.5 0.6 0.70

0.1

0.2

0.3

0.4

0.5

0.6

0.7

20

30

40

10 2020

2222

8

0

14

0 0.1 0.2 0.3 0.4 0.5 0.6 0.70

0.1

0.2

0.3

0.4

0.5

0.6

0.7

10

Reh=80000

Percentage of base-pressure increase (Reh=40000, 80000)

Reh=40000

ly/h

Optimum size : ly/h=0.15 ~ 0.35, lz/h=0.15 ~ 0.35

20 30 40

1

2

3

1 2 3

ly/θly/δ

10 20 401 2 3

30

10

20

40

1

2

3

30

lz/hlz/θlz/δ

lz/hlz/θlz/δ

ly/hly/θly/δ


Research

Cpb

How about two-dimensional fence?

z/W

Flow

O without disrupter

□ ly/h = 0.25, lz/h = 0.2

△ ly/h = 0.25, lz/h = 5

Reh=20000

-2

-1.5

-1

-0.5

0

-0.5 0 0.5

• Introduction of three-dimensional motion is essential for drag reduction.


Research

Four sizes of wake disrupter(lx / h, ly / h, lz / h) = (0.2,0.2,0.2) (lx / h, ly / h, lz / h) = (0.4,0.2,0.2) (lx / h, ly / h, lz / h) = (0.2,0.4,0.2) (lx / h, ly / h, lz / h) = (0.2,0.2,0.4)

Large eddy simulation at Reh=4200

δ / h =1.4, θ / h = 0.186

Domain: 15/3 ≤≤− hx29/29 ≤≤− hy4/0 ≤≤ hz

670/ == ∞ νθθ uRe

Grid points:)(64)(240)(350 zyx ××

4200/ == ∞ νhuReh

)(4)(18)(14 zyx ××Grid points inside a wake disrupter:

Immersed boundary method

(Kim, Kim & Choi 2001)

-2 2

0//, =∂∂=∂∂= ∞ ywyvuu

Convectiveoutflow b.c.

x

y

z

Turbulentboundary

layer(Reθ =670)

h

0//, =∂∂=∂∂= ∞ ywyvuu

∞u

∞u

Model vehicle(Tombazis &Bearman 1997)

Turbulentboundary

layer(Reθ =670)


Research

Without disrupter

④①②③

①

②

③

④

Base pressure along the spanwise direction

(lx / h, ly / h, lz / h) = (0.2,0.2,0.2)

(lx / h, ly / h, lz / h) = (0.2,0.2,0.4)

(lx / h, ly / h, lz / h) = (0.2,0.4,0.2)

(lx / h, ly / h, lz / h) = (0.4,0.2,0.2)

1 2-1-2

Maximum base-pressure recovery is about 16% in spite of the differences in θ and Reh with the experiment.

⇒ Wake disrupter is an effective control method applicable to various Reynolds numbers.


Research

With disrupter(lx,ly,lz)=(0.2,0.2,0.2)x

z

Flow

Wake structures

Without disrupter

• Iso-surfaces of pressure

x

z

Flow

Large-scale, two-dimensional Karman vortex cores exist.

Kármán vortex cores are broken up into smaller-scale vortices and the wake becomes completely three-dimensional.


Research

Wake structures

(lx / h, ly / h, lz / h) = (0.2,0.4,0.2)

base-pressure recovery: 12%

(lx / h, ly / h, lz / h) = (0.2,0.2,0.4)

base-pressure recovery: 16%

Uncontrolled flow

- Wake disrupter substantially suppresses vortex shedding.

- The naturally occurring vortices right behind the body certainly disappear with disrupters.


Research

Conclusions

• In the present study, we considered two different bluff bodies; a circular cylinder and a two-dimensional model vehicle

• The distributed forcing (active control) produced a significant amount of drag reduction for both bodies at quite different Reynolds numbers, and the drag reduction was achieved by direct modification of vortical structures in the wake rather than separation delay.

• The wake disrupter (passive device) also produced a significant amount of drag reduction for the model vehicle (the same result is also expected for the circular cylinder).

• Both the distributed forcing and wake disrupter introduced the spanwise mismatch in the vortex shedding process (either phase difference or frequency difference);Overall vortex shedding was weakened due to the spanwise incoherency;Then, the base pressure was increased and drag was reduced.


Research

N

S- -+ + -

How to realize the distributed forcing from EM tiles?

Top view

One wavelength of forcing

U

CylinderOr massively separating flow

such as an airfoil at a large angle of attack!

Documents

Active Control of Flow over a Circular CylinderTurbulence and Flow Control Seoul National University Center for Research Contents 1. Characteristics of flow over a circular cylinder