UNRAVELING

TURBULENCE
A.E.P. Veldman and R.W.C.P. Verstappen
University of Groningen
 Panta Rhei
Heraclitus (535-475 BC)
 Navier-Stokes equations
Claude Navier (1822) George Stokes (1845)

• Conservation of momentum (incompressible)

∂u 1
+ |(u · grad)u = − grad p + |div(ν grad u)}
| ∂t ρ {z
{z } {z
{z }
convection | } diffusion
evolution pressure
non-linear ν very small
creator of turbulence

convection U L
• ratio ∼ = Re(ynolds number)
diffusion ν

Osborne Reynolds (1894)

 Generation of turbulence
∂u ∂u ∂ 2u
+ u + ··· = ··· + ν 2
∂t ∂x ∂x
∂u
u ∼ sin ωx ⇒ u ∼ ω sin ωx cos ωx
∂x
∂u
⇒ ∼ − 21 ω sin 2ωx − ν ω 2 sin ωx
∂t
Energy cascade:
Frequency doubling until diffusion comes into action.
Big whorls have little whorls,
Which feed on their velocity,
And little whorls have lesser whorls,
And so on to viscosity.
(L.F. Richardson, 1881–1953)
 Scaling and complexity
• Kolmogorov (1941) scaling
dissipation of energy ǫ = de/dt [L2T −3]
kinematic viscosity ν [L2T −1]
ℓ ∼ (ν 3/ǫ)1/4; τ ∼ (ν/ǫ)1/2

size small eddies ℓ ∼ Re−3/4 × size large eddies L

time scale small eddies τ ∼ Re−1/2 × time scale large eddies T

• Complexity of flow
(Re3/4)3 × Re1/2 = Re11/4
• Re 10× larger ⇒ complexity 1000× larger!
 Leonardo’s turbulence

Eddies in water
(≈ 1507)
 Laminar vs. turbulent

laminar flow (forced) turbulent flow

Turbulent flow is better able to follow curved walls.

Reduces size of wake, and herewith drag!
 Exploiting turbulence
Drag reduction by
delaying separation
golf ball

speed skater
head torso
& legs
 Modelling turbulence
mean
RaNS Reynolds-averaged Navier-Stokes:
model all turbulence (steady)

LES Large-eddy simulation (unsteady):

resolve large eddies; flow behind a square cylinder
model smaller eddies
snapshot

DNS direct numerical simulation:

resolve all flow structures
 Modelling turbulence (2)
Prediction Spalart (2000):
airplane Re = 107−8

model RaNS LES DNS

grid points 107 1011.5 1016
memory (bytes) 1010 1014.5 1018.5

time steps 103 106.7 107.7

flops 1014 1022 1027

speed (flop/s) 109 1017 1022

year 1990 2045 2080

But: Algorithmic improvements bring future sooner!

Square cylinder
at Re=22.000
Testcase RaNS and LES - 1997
 Attempts by RaNS and LES
Models were tuned to known answer

DNS is pure Navier–Stokes – and similar grid ...

 Challenges
• Ten years ago Re = 104 was not really possible.
So how about the following applications?
medium speed Reynolds
cyclist (tourist) air 20 km/h 1 · 105
golfball (pro) air 250 km/h 2 · 105
speed skater (pro) air 45 km/h 5 · 105
swimmer (pro) water 5 km/h 3 · 106
car (Dutch) air 80 km/h 5 · 106
shark water 20 km/h 2 · 107
airplane air 900 km/h 3 · 107
ship water 20 km/h upto 109
Note: air is 15× more viscous than water

• A ship (+screw) is even 1015 more expensive!

 Boosting performance
• Computer designers
computer performance: 30× per decade
• Numerical mathematicians
algorithm performance: 30× per decade

Together: factor 1000 per decade!

⇒
i.e. factor 10 in Reynolds number

NB: Similar progress is found in other computational disciplines.

 Numerical progress
Discretization of dφ/dx

Lagrangian interpolation Symmetry preservation

φ0 φ0

φ+ φ+
φ- φ-
h- h+ h- h+
111111111111111
000000000000000 111111111111111
000000000000000
000000000000000
111111111111111 000000000000000
111111111111111
x- x0 x+ x- x0 x+
dφ φ+ − φ−
=
dφ h2− φ+ + (h2+ − h2−)φ0 − h2+ φ− dx h+ + h−
=
dx h+h−(h+ + h−) Skew-symmetric expression:
Coefficient of φ0 can make sys- – system never singular!
tem singular! – no artifical diffusion!
 Balance at smallest scales

• Turbulence is subtle balance between production by

(non-linear) convection and destruction by diffusion

• Numerical diffusion must not interfere with

this balance !!

• Skew-symmetric discretization has no numerical diffusion

• Similar story for skew-symmetric turbulence modelling

 Evolution of energy
dφh
+ Lhφh = 0
dt

• ‘Energy’ ||φh||2h = φ∗hHφh evolves in time as

d
||φh||2h = −(Lhφh)∗Hφh − φ∗hHLhφh
dt
= −φ∗h(HLh + (HLh)∗)φh
(H represents local grid size)
⇒ – energy is conserved iff HLh is skew-symmetric
– energy dissipates iff symmetric part of HLh is
positive definite
Channel turbulence (1)

20

u+

16

u += y +

12

u+= 2.5 ln y ++ 5

4 DNS 4th−order 64x64x32

Kim et al (1987)
Kuroda et al (1995)
Gilbert & Kleiser (1991)
0
1 2 5 10 20 50 100 y + 200

Calculated mean flow

 Channel turbulence (2)
Grid coarsening: 128 → 16 across channel
Turbulent statistics u′u′
3

urms

DNS 4th-order 64x96x32

DNS 4th-order 64x64x32
DNS 4th-order 64x32x32
1 DNS 4th-order 64x16x32
DNS 2nd-order 64x64x32
Experiment Kreplin & Eckelman (1979)
DNS Kim et al. (1987)

0
0 10 20 30 y+ 40

Kim et al. (1987) 128 points; now reduced to 32 or even 16!

Algorithmic gain 1000, i.e. 20 years!
 Current ‘price’ for DNS
golf ball car ship
skater airplane
Reynolds 105 107 109
grid points 1010 1014.5 1019
memory (bytes) 1013 1017.5 1022
time steps 105 106 107
flops 1018 1023.5 1029
performance (flop/s)∗ 1012 1017.5 1023
∗
2-week run

Enough room left for algorithmic and modelling

improvement ... ;-)
 Price tag for DNS
2-wk run without further numerics progress
speed
memory ship
Zeta
flops/sec & bytes

Exa airplane

car
Peta swimmer

golf ball
Tera

2007
Giga 4 5 6 7 8 9
10 10 10 10 10 10
Reynolds number
 Epilogue
Sir Horace Lamb in 1932 (then aged 83) stated:
I am an old man now, and when I die and
go to heaven there are two matters on which
I hope for enlightenment. One is quantum-
electrodynamics, and the other is the turbulent
motion of fluids. And about the former I am
rather optimistic.
Computer simulation may shed some light on the latter.