Turbulent Rayleigh-Benard Convection
Sponsored Links
This presentation is the property of its rightful owner.
1 / 49

Turbulent Rayleigh-Benard Convection A Progress Report PowerPoint PPT Presentation


  • 90 Views
  • Uploaded on
  • Presentation posted in: General

Turbulent Rayleigh-Benard Convection A Progress Report. Work done in collaboration with Eric Brown, Denis Funfschilling And Alexey Nikolaenko, supported by the US Department of Energy. Guenter Ahlers Department of Physics and iQUEST

Download Presentation

Turbulent Rayleigh-Benard Convection A Progress Report

An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -

Presentation Transcript


Turbulent Rayleigh-Benard Convection

A Progress Report

Work done in collaboration with

Eric Brown, Denis Funfschilling

And Alexey Nikolaenko,

supported by the US Department of Energy

Guenter Ahlers

Department of Physics and iQUEST

UCSB


Rayleigh-Benard Convection Cell

Fluid: H2O

Wall: Plexigl.

G = D/L = 1.0

D

Q

Three samples:

Small: D = L = 9 cm

Medium: D = L = 25 cm

Large: D = L = 50 cm

DT = 20 oC : R = 1011

Q = 1500 W

L

DT

R = a g L3 DT / k n

Q

Prandtl No. s = n/k = 4.4

( H2O, 40 oC )


Why is it interesting?

Important process in the

Atmosphere:Weather

Mantle:Continental Drift

Outer core:Magnetic field

Sun:Surface temperature

etc.:???

Interesting Physics


Models


A central prediction: Heat transport.

Define Nusselt number N = leff / l ;leff= Q / ( DT / L )

Various models were proposed:

Malkus (1954), Priestly (1959): N ~ R1/3

Kraichnan (1962): N ~ R1/2

Castaing et al. (1989)

Shraiman + Siggia (1990): N ~ R2/7

Grossmann and Lohse (2000):

No single power law; Crossover between

two power laws


C.H.B. Priestley

[Quarterly Journal of

the Royal

Meteorological

Society 85, 415 (1959)]

Define N = leff / l ;leff= Q / ( DT / L )

Q = N l DT / L = heat-current density

Assume that there are power laws and that the

R- and s-dependence of N separates:

N = f( s ) Rg ;R = (a g DT / k n) L3

Q = f( s ) Rg l DT / L

Q = f( s ) (a g DT / k n)gl DT L(3g - 1)

Assume that the heat-current density Q is determined by the BLs and does not depend on the distance between them. Then

3 g - 1 = 0; g = 1/3; N ~ R1/3


R. Krishnamurty and L.N. Howard, Proc. Nat. Acad. Sci. 78, 1981(1981):

Large Scale Circulation (“Wind of Turbulence”)

R = 6.8x108

s = 596

  • = 1

    cylindrical

    slightly tilted

    in real time

Movie from the group of K.-Q Xia, Chinese Univ., Hong Kong

connects the BLs and invalidates the simple models.


Q

Q

T1

thermal boundary layer

cold plumes

schematic drawing

of the flow structure

“wind” or

large scale

circulation

( T1 + T2 ) / 2

thermal boundary

layer

hot plumes

T2 > T1

See, e.g., X.-L. Qiu and P. Tong, Phys. Rev. E 66, 026308 (2002)


However, assuming no interaction between

BLs is not needed to get 1/3 !

W.V.R. Malkus, Proc. Roy. Soc. (London) A 225, 196 (1954)

Assume laminar BLs with conductivity l

and DT/2 across each:

Q = N l DT / L = l (DT/2) / l

l = BL “thickness”

l = L / 2N

Assume laminar BLs are marginally stable:

R = Rc = a g l3 DTc / k n = O(103); DTc =DT/2

l ~ (DT/2)-b ~ R-b; b = 1/3; N = L/2 l;

N ~ R1/3

z / l

Wind

direction

x

Experiment: b not = 1/3 and depends on horizontal location !

S.-L. Lui and K.-Q. Xia, Phys. Rev. E 57, 5494 (1998). s = 7.


1975: D.C. Threlfall, J. Fluid Mech. 67, 17 (1975).

1987: F.Heslot, B. Castaing, + A. Libchaber(Chicago), Phys. Rev. A 36, 5870 (1987).

N ~ R0.282

Mixing layer model (bulk, BL, and plume region between them)

ofthe Chicago group

[Castaing, Gunaratne, Heslot, Kadanoff, Libchaber, Thome, Wu, Zaleski, & Zanetti,

J. Fluid Mech. 204, 1 (1989)]

and of

B.I. Shraiman and E. Siggia, Phys. Rev. A 42, 3650 (1990):

N ~ s-1/7 R2/7; 2/7 = 0.2857… .


S. Grossmann and D. Lohse, J. Fluid Mech. 407, 27 (2000) (GL)

start with the kinetic and thermal dissipation rates

Their volume averages follow from the Boussinesq equations and are given by

GL set each equal to a sum of a BL and a bulk contribution:

They assume that the separate contributions can be modeled using approximations

to the length, temperature, and time scales, e.g.

(assumes laminar BLs, uniform in the

x-y plane, with conductivity l)

(based on Blasius BL model)

etc.


log( s )

log( R )

No simple power laws, but rather cross-overs from a small-R to a large-R

asymptotic region.

Various regions in the R - s plane, depending on which dissipative

term dominates, etc. For s > 1 and large R, IVu pertains.

There eu and eq are both bulk dominated.


yield

and

At large R

else


D. Funfschilling, E. Brown, A. Nikolaenko, and G. A., J. Fluid Mech., submitted.


1.) No power law

2.) 4 parameters of the

GL model were

determined from a fit

to these data

X. Xu, K.M.S. Bajaj, and G. A., Phys. Rev. Lett. 84, 4357 (2000);

G. A. + X. Xu, Phys. Rev. Lett. 86, 3320 (2001)


A.) The important components have been identified:

1.) top and bottom boundary layers

2.) “plumes”

3.) large-scale circulation

B.) The nature of the interactions between boundary layers,

plumes, and large scale circulation is not so clear.

C.) The GL model can be fitted to existing Nusselt data

by adjusting its four undetermined coefficients

D.) Adjustment of a fifth parameter gives reasonably

good agreement with the measured Reynolds numbers

of the LSC.


New Nusselt-Number

Measurements


N / R1/4

Prandtl Number s

R = 1.8x109

s -1/7

R = 1.8x107

X. Xu, K.M.S. Bajaj, and G. A., Phys. Rev. Lett. 84, 4357 (2000);

G. A. + X. Xu, Phys. Rev. Lett. 86, 3320 (2001)

K.-Q. Xia, S. Lam, and S.-Q. Zhou, Phys. Rev. Lett. 88, 064501 (2002).

S. Grossmann and D. Lohse, Phys. Rev. Lett. 86, 3316 (2001).


Foam inside of here

Water cooled Cu top plate

adiabatic side shield

Plexiglas side wall

Joule heated Cu bottom plate

Adiabatic bottom-plate shield

Leveling and support plate

Catch basis

E. Brown, A. Nikolaenko, D. Dunfschiling, and G.A., Phys. Fluids, submitted.


D. Funfschilling, E. Brown, A. Nikolaenko, and G. A., J. Fluid Mech., submitted.


2 %

D. Funfschilling, E. Brown, A. Nikolaenko, and G. A., J. Fluid Mech., submitted.


  • The GL model can not reproduce the effective exponent

  • = 0.333 ~ 1/3 derived from the Nusselt number data

    for R > 1010.


Reynolds-Number

Measurements


0.63 cm


3

4

2

5

1

5

6

7

0

1

q

0

6

7


C00

-C04


Re = (L / t1)(L / n)

Medium Sample

X.-L. Qiu and P. Tong,

Phys. Rev. E 66, 026308 (2002).

unpublished

Large Sample


A.) The important components have been identified:

1.) top and bottom boundary layers

2.) “plumes”

3.) large-scale circulation

B.) The nature of the interactions between boundary layers,

plumes, and large scale circulation is not so clear.

C.) Models yielding relationships between, the Nusselt number,

Rayleigh number, Prandtl number, and Reynolds number

(of the LSC) are at best good approximations, but for

large R miss important physics.


LSC Reversals

E. Brown, A. Nikolaenko, and G. A., Phys. Rev. Lett., submitted.


Ti = <T> + d cos( ip/4 + q )


Rotation


Cessation


probability distribution of |Dq| for reorientations with d/<d> < 0.25


A.) LSC “reversal” can occur via

1.) rotation of the vorticity vector (“rotation)

2.) shrinking of the vorticity vector, followed by

re-development with a new orientation (“cessation”)

B.) Cessation is followed by re-development of the LSC

in a circulation plane with an arbitrary new orientation,

i.e. P(Dq) ~ constant.

C.) Rotation through an angle Dq has a powerlaw

probability distribution P(Dq) ~ Dq-g with g ~ 4.

D.) Reversals are Poisson distributed.


More LSC

Dynamics

D. Funfschilling and G. A., Phys. Rev. Lett. 92, 194502 (2004).


Shadowgraph

lens

Hot plumes/rolls

near the bottom plate

appear dark

Cold plumes/rolls

Near the top plate

Appear bright

pin hole

and LED ligh

source

beam-splitter

lens

Rayleigh

Bénard cell

2o tilt

mirror


dt = 0.0s

34

Correlation Functions:

0.3s

dt = 0.0s

0.6s

0.9s

37

1.5s

1.2s

dt = 0.9s


The maximum of the correlation

Function is located at DX, DY

Relative to its origin (center).

Viewed from

Above:

direction of plume

movement and

presumed direction of

circulation of LSF

Lowest point

Speed

Q

Direction of 2 deg. tilt

Highest point


The angle qof the plane of the large-scale-flow circulation,

and its time correlation function

R = 7.0 x 108


Assumption:

Near the center of the top and bottom plate plumes/rolls

follow the large-scale flow

Conclusion

Near the center of the top and bottom plate the large-scale flow direction oscillates about the vertical axis of the cell.

This oscillation has the same frequency as the periodic

signals seen by others in measurements at individual points.

The frequency yields a Reynolds number consistent with measurements by other methods and the GL model.


  • Login