Rotating Taylor Green Flows

6 downloads 67 Views 1MB Size Report
Sep 26, 2013 ... How does 2D inverse cascade saturate? For Ro ≪ 1 what is Re ≫ 1? 2 / 1. Alexandros Alexakis, Martin Schrinner. Rotating Taylor Green ...
Rotating Taylor Green Flows Alexandros Alexakis, Martin Schrinner

September 26, 2013

1/1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Rotating Flows A rotating world Stars, planets ... Earth’s atmosphere ... laboratory experiments ...

Weak Wave turbulence, is it there? What controls forward to backward cascade ratio? How does 2D inverse cascade saturate? For Ro  1 what is Re  1? 2/1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Taylor Green Forcing: A test case ∂t u + u · ∇u + 2Ω × u = −∇P + ν∇2 u + F

Ω f q=2

1 • F= Lz

Z Fdz = 0,

F = F0

 

ex sin(qx) cos(qy ) sin(qz) − ey cos(qx) sin(qy ) sin(qz)  ez 0

1 • hF · ∇ × Fi = V

Z F · ∇ × FdV = 0

• u = F is not a solution of the Euler equations 3/1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Control Parameters

Re = UL/ν, U = kuk,

U=

Ro = U/2LΩ

p F0 L,

U = 1/3 L1/3

RoF =

FL 2ΩL

ReF =

√ FL3 ν

RoU =

kuk 2ΩL

ReU =

kukL ν

RoD =

1/3 2ΩL2/3

ReD =

1/3 L4/3 ν



Ω f q=2

4/1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Parameter Space

••• Slow rotating 5/1

•••

Alexandros Alexakis, Martin Schrinner

•••

•••

Rotating Taylor Green

••• Fast rotating

The Runs

167 Runs Resolutions 643 − 5123 0 < ReD . 700 0.0...01 < RoD < ∞

6/1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Different behaviors observed Energy evolution

“Laminar” Bursts

Quasi-2D Inverse cascade Quasi-isotropic turbulence

7/1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Parameter Space

8/1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Parameter Space

9/1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Expansion

Ro  1 u → = RoF u, τ = RoF t and ReF−1 = λRoFn+1

∂t u + ez × u − ∇P = F − RoF

  ∂τ u + w × u − ∇P 0 + λRoFn ∇2 u. or

∂t u + L [u] = F + Ro P [−∂τ u + u × w] + λRo n ∇2 u P[u] = projection to incompressible fields P [ez × g]

L[g] =

u = u0 + Ro u1 + Ro2 u + . . . 10 / 1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

What we can show

For n = 0 O(ReF ) = O(RoF−1 ) Laminar solution, kuk ∼ F0 /Ω, (ReU = RoF ReF ) u=0 Nonlinearly (energy) stable for ReF < C1 RoF−1 . For n = 1 O(ReF ) = O(RoF−2 ) u=0 Laminar solution, kuk ∼ F0 /Ω, −2 Linearly stable for ReF < C2 RoF . For n = 3 O(ReF ) = O(RoF−4 ) Laminar solution, kuk ∼ F0 /Ω, u 6= 0 −4 2D-flow-unstable for ReF > C4 RoF . For n > 3 O(Re) > O(Ro −4 ) No Laminar solution exists.

11 / 1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Parameter Space

12 / 1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Energy dissipation Have we reached high enough Re yet?

13 / 1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Rotating Taylor Green quasi 2D flows

vertical vorticity

14 / 1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green

Conclusions What is high Re for low Ro is not a trivial question. Do not extrapolate present results to other forcing functions (Different behavior for F 6= 0 is expected.)

15 / 1

Alexandros Alexakis, Martin Schrinner

Rotating Taylor Green