You are on page 1of 28

Turbulence in Astrophysics

(Theory)
Wolfram Schmidt
Institut für theoretische Physik und Astrophysik Universität Würzburg

Stirring of Fluid
‡ Mechanical force stirring fluid into rotational motion ‡ Turn-over time T, wavelength L ‡ What happens in the limit t ’? It depends on the Reynolds number!

LV 1 L Re ! ! R R T
21 September 2004 Astrophysical Turbulence

2

2

Laminar vs. Turbulent Flow
Reynolds, 1883

‡ If Re is relatively small, only eddies of size L are produced ‡ For Re ~ 1000, the motion of adjacent fluid layers becomes unstable

21 September 2004

Astrophysical Turbulence

3

21 September 2004

Astrophysical Turbulence

4

Fluid motion forces vortices to stretch, and a stretching vortex must fold to accomodate an increasing length in a fixed volume. To the extent that the flow is scaling, I conjecture the vortex tends toward a fractal.
Mandelbrot, The Fractal Geometry of Nature

21 September 2004

Astrophysical Turbulence

5

Vortices

Turbulent fluid motion is inherently rotational
21 September 2004 Astrophysical Turbulence 6

Strain and Vorticity
Symmetric derivative Antisymmetric derivative

Sik !

1 2

x i vk  x k vi

Wik !

1 2

x i vk  x k vi

Rate of strain

S ! 2 Sik Sik
Dilatation

Vorticity

! 2WikWik ! “ v v
Astrophysical Turbulence 7

d ! Sii ! “ ™ v
21 September 2004

Vortex Formation

Vortices are streched and folded in three dimensions
21 September 2004 Astrophysical Turbulence 8

Porter et al. ASCI, 1997

The Turbulence Cascade
Richardson, 1922; Onsager, 1945

‡ Breaking up of laminar flow structure due to large |S| produces high vorticity ‡ Force of wavelength L produces structure on scales much smaller than L for high Re ‡ Small vortices are quasi random Turbulence is a non-linear multi-scale phenomenon
21 September 2004 Astrophysical Turbulence 9

Isotropic Turbulence
Taylor, 1935

‡ Statistically, there is no prefered direction (random orientation of vortices) ‡ In nature, turbulence is never exactly isotropic on large scales (forcing, BCs) ‡ However, turbulent flows tend to become asymptotically isotropic towards small scales (randomisation by non-linear energy transfer)
21 September 2004 Astrophysical Turbulence 10

The Kolmogorov Theory
‡ Hypothesis of local isotropy: At sufficiently high
Re, the dynamics on small scales tends to become
statistically isotropic

‡ First similarity hypothesis: The statistics of
isotropic velocity fluctuations on sufficiently small scales are universal und uniquely determined by the viscosity and the rate of kintetic energy dissipation ‡ Second similarity hypothesis: There is a subrange of scales for which the statistics of turbulent fluid motions are independent of the mechanism and the length scale of dissipation
21 September 2004 Astrophysical Turbulence 11

The 5/3 Power Law (K41)
log E

E (k ) ! C

2/3

k 

5 / 3

k -5/3

Rate of dissipation Wave number k = 2 /l Length scale of viscous dissipation
K

L

-1

-1 K

log k

L
K

} Re

3/ 4

!

3 / 4 1 / 4
Astrophysical Turbulence

21 September 2004

12

But the hope that Ähomogeneous turbulence³ would be a sensible model was dashed by Landau & Lifschitz 195319531959, which notes that some regions are marked by very high dissipation, while other regions seem by contrast nearly free of dissipation.
Mandelbrot, The Fractal Geometry of Nature

21 September 2004

Astrophysical Turbulence

13

Realistic Turbulence
‡ Convective boundary layers: Anisotropy in
stratified medium (convection zones, atmospheres)

‡ Turbulent combustion: Anisotropy across flame
surface, transient flow (thermonuclear supernovae)

‡ Gravoturbulence: Inhomogeneous and supersonic
turbulence in self-gravitating fluids (star formation)

‡ MHD turbulence: Instability of fluid motion due to
interaction with magnetic field, multi-scale anisotropy (ionized gas in ISM, jets, accretion disks)
21 September 2004 Astrophysical Turbulence 14

The Navier-Stokes Equation NavierConservation of momentum

D v ! “P  f  “ ™ Dt
Lagrangian time derivative Mechanical, magnetic, gravitational forces Viscous dissipation tensor
ik

D x !  v ™“ Dt xt
21 September 2004

!2 

Sik

!2

Sik  1 ik 3
15

Non-linear advection
Astrophysical Turbulence

Further Equations
Conservation of energy

D e  “ ™ Pv  ™ v  c P “T ! f ™ v Dt
Mass conservation Poisson equation

D Dt 

“™v ! 0

(J ! 4TG
Maxwell equations in the case of MHD

21 September 2004

Astrophysical Turbulence

16

Statistical Theories
‡ Mixing length theory: one characteristic length -peak) scale lM = HP (Kolmogorv spectrum ‡ ODT models: one-dimensional stochastic process
for eddy size (reproduces Kolmogorv spectrum)

‡ PDF models: determine probability distributions
for velocity fluctuations etc.

‡ Reynolds stress models: dynamical equations
for moments of fluctuating fields
21 September 2004 Astrophysical Turbulence 17

Stellar Convection
‡ Full Reynolds stress model for compressible turbulence (Canuto, 1997): multitude of coupled, non-linear PDEs hopeless ‡ Feasible model: reduced set of eqns. for mean K , Fc , and average squared fluctuations of temperature and horizontal velocity (Kupka, 1999)
Closure relations for higher order moments Non-local & anisotropic
21 September 2004 Astrophysical Turbulence 18

Stellar Convection: Convective Flux

21 September 2004

Astrophysical Turbulence

Kupka MPA, 2004
19

Stellar Convection: Vertical RMS Velocity

21 September 2004

Astrophysical Turbulence

Kupka MPA, 2004
20

Numerical Simulations
‡ Direct numerical simulation (DNS): Static grid, NSE or numerical viscosity ‡ Large Eddy Simulation (LES): Subgrid scale model for unresolved turbulence ‡ Smooth particle hydrodynamics (SPH): Particle ensemble represents the flow ‡ Adaptive mesh refinement (AMR): Hierarchy of dynamically generated grid patches
21 September 2004 Astrophysical Turbulence 21

Thermonuclear Supernovae
‡ Runaway turbulent deflagration of C+O in a Chandrasekhar-mass white dwarf ‡ PPM for hydrodynamics (Fryxell et al., 1989) ‡ Subgrid scale model for turbulent flame speed (Niemeyer & Hillebrandt, 1995) ‡ Level set method for flame surface tracking (Reinecke et al., 1999 ) ‡ Homologous grid expansion to follow the explosion (Röpke, 2004)
21 September 2004 Astrophysical Turbulence 22

t=0s

History of a SN Ia Explosion

t = 0.3 s

t = 0.6 s

Röpke et al. MPA, 2004
t=2s
21 September 2004 Astrophysical Turbulence 23

21 September 2004

Astrophysical Turbulence

24

Turbulence in the ISM
‡ Supersonic turbulence in self-gravitating gas ‡ Thermal processes, magnetic fields ‡ Paradigm of turbulent star formation: Turbulence can induce local gravitational collapse, albeit it provides global support ‡ SPH treatment: e.g. Klessen, 2001 ‡ AMR treatment with PPM/ZEUS: e.g. Kritsuk & Norman, 2002; Abel et al., 2002
21 September 2004 Astrophysical Turbulence 25

21 September 2004

Astrophysical Turbulence

26

21 September 2004

Astrophysical Turbulence

27

Résumé
‡ Significant developments in the treatment of turbulent convection via statistical models ‡ Three-dimensional simulations with sophisticated codes running on extremely powerful computers offer exciting insights ‡ However, most simulations are ignorant of small-scale turbulence (SGS models!) ‡ AMR is excellent for inhomogeneous and transient and astrophysical flows, but is it appropriate for turbulence?
21 September 2004 Astrophysical Turbulence 28