Multi-scale :

Introduction to turbulence
Computational Fluid Dynamics

* : Covered partly in the exam.

Primary references :
• Pope, Turbulent flows
• Van Dyke, An Album of Fluid Motions
• John Kim, Class material for "Introduction to Turbulent Flows"
 Turbulent flows from simulations
Examples from my portfolio
Direct numerical simulation, Large-eddy simulation
Theoretical field or practical field?
 Turbulence is always around us
Air flow around air purifier

Source:
Wonjung Kim, Sogang Univ.
 Laminar vs. turbulent flows
Flow around a circular cylinder
Features of turbulent flows

Flow inside a centrifugal pump

Turbulent flows need special approach for accurate CFD prediction

Turbulence modeling
 Features of turbulent flows
Reynolds’ classical experiment (1883)
For sufficiently large Reynolds number, flow becomes turbulent
Random & chaotic

Source: Van Dyke, Album of Fluid Motions

 Onset of turbulence
Re ~ 1000

Turbulent

Laminar

Source: Van Dyke, Album of Fluid Motions

 Onset of turbulence
Horizontal screen + acceleration

Source: J. Kim, Class material for "Introduction to Turbulent Flows"

 Onset of turbulence
Higher Re

Source: J. Kim, Class material for "Introduction to Turbulent Flows"

 Reynolds number comparison
Turbulent jets from circular nozzle

Source: Van Dyke, Album of Fluid Motions

 Reynolds number comparison
Wake behind a plate at   4300 and a grounded ship at   10
For high Re, flow solution becomes independent of Re and dependent
on the geometry

Source: Van Dyke, Album of Fluid Motions

 Onset of turbulence
Turbulence starting from vortices - Karman vortex shedding

Re=140

Re=10,000 Source: Van Dyke, Album of Fluid Motions

 Onset of turbulence
Karman vortex shedding

Source: http:/ / viruns.egloos.com/ m/ 5112035 Source: J. Kim, Class material for "Introduction to Turbulent Flows"
 Onset of turbulence
Vortex visualization around F-18
smooth

helical

chaotic
& random

http:/ / commons.wikimedia.org/ wiki/ File:F- 18_model_in_water_tunnel_with_dye_injections_(side_view).jpg

http:/ / commons.wikimedia.org/ wiki/ File:FA- 18_model_water_tunnel_visualisation_DFRC_763x1280.jpg
 Onset of turbulence
Kelvin-Helmholtz (KH) instability (to small scales)
Two streams with different speeds => Flow roll-up develops

low s

x
fast

http:/ / www.brockmann- consult.de/ CloudStructures/ kelvin- helmholtz- instability- description.htm

Further to even smaller scales

A hint to energy cascade

Fontane et al. (2008)

Buoyancy & density-driven turbulence
Wide range of length scales

Source: J. Kim, Class material for "Introduction to Turbulent Flows"

Source: Daum
 Scales of turbulence
Small-scale coherent structures near a wall
Larger wall area leads to smaller drag?
Riblets push away (similar size) streamwise vortices

철 lbdiameter
103 184m
~
increase
drag

Fastskin from Speedo

Scientific American

Source: “ Turbulent flows” by Pope

 Scales of turbulence
Atmospheric turbulence in Jupiter (~ 10,000 km)

Flow information: Jupiter: The Planet, Satellites and Magnetosphere By Fran Bagenal
Source: J. Kim, Class material for "Introduction to Turbulent Flows"
 Scales of turbulence
Non-premixed turbulent flame : Turbulent mixing + Chemical reaction

Source: J. Kim, Class material for "Introduction to Turbulent Flows"

 Principal observations
Instability develops when it can not be suppressed by viscosity
as Relx
Random - especially small scales

Three-dimensional

Wide range of scales

Vorticity

Mixing

Intermittent

Scale-similar - especially in jets

Van Dyke, Album of Fluid Motions
Large-scale features are dependent on geometry but independent on Re
Smaller scales depend on Re
A rough estimate of turbulent time scale
Compare time scales of molecular and turbulent mixing (heat transfer)
Heat transfer in a room of size  without flow: *scale
analysis
  Θ Diffusion rough
④
Θ estimate
 

C
 

ΔΘ
Ss
Scaling with temperature difference across the room ΔΘ
ΔΘ 
, ~  ⇒  !""#$ "!  ~
  
Turbulent mixing time with turbulent scales proportional to  , ′ ㅅ

 velocityvortex
 ~ Assume  % 
′
Ratio of turbulent to molecular mixing time
V
   & 1 8

10001 실   ~
 ′  ′ 
'( )  1* Source: “ Turbulent flows” by Pope

Re can be interpreted as the ratio of molecular to turbulent time scales

As Re ↑,  ≪  : Turbulent mixing is faster
 Mean and turbulent kinetic energy
To understand scale interaction

Reynolds decomposition :  ,    ,  + , , 
까

)
3
Ensemble average :  ,   lim 0 ∑0
563  '5*
, 
0→2

Momentum equation for incompressible flows

789 7 3 7> 7 ? 89
+  5  < = 7: + & 7: 7:

CG
7 7:; 9 ; ;

Equations for mean and fluctuation components

789 7 3 7> 7 ? 89 789@ 8;@
7 7
+ 7:;
 5  < = 7: + & 7: 7: < 7:;
X 고
9 ; ;

789@ 7 3 7>@ 7? 89@

+  5, + , 5 + 5, , < 5, ,  < = 7: + & 7: XUi
7 7:; 9 ; 7:;

A
789 7 ? 89 89 7B
Multiply by 5 and 5, , and rearrange 5  =
7 7 7
 Mean and turbulent kinetic energy
3
Equation for mean kinetic energy (MKE, C  5 5 )

.
large scale
789 7 3 7> 7 ? 89 789@ 8;@
7
+ 7:;
 5  < = 7: + & 7: 7: < 7:;
9 ; ;

C C  1 5
+  + 5 5, , +  ) < 2&5 K5  5, , < 2&K5 K5
   I 
간
트
D FGH L M>
마
3 , , ismal
Equation for turbulent kinetic energy (TKE, D ≡ ) scale
 5 5 RHS : source or
789@ 7 3 7>@ 7? 89@
+ 7:  5, + , 5 + 5, , < 5, ,  < = 7: + & 7: sink of C and D
7 ; 9 ; 7:;

6 D

+ 
D
+
 1 , , , 1 , ,
  2
, , , ,
 5 5 +  ) < 2&5 K5  <5 
I
5

, ,
< 2&K5 K5
d O
Da FGH L N>
권
789
They exchange energy via TKE production ()  <5, , )
7:;
 Turbulent kinetic energy production
Production of TKE is usually positive
TKE balance (budget) in Turbulent boundary layer

Wall

Source: “ Turbulent flows” by Pope

 Turbulent kinetic energy production
Production of TKE is usually positive
TKE balance (budget) in Turbulent round jet

Source: “ Turbulent flows” by Pope

 Turbulent kinetic energy production
Production of TKE is usually positive
7B 7B 7 3 789
+  + 5 5, , +  ) < 2&5 K5  5, , < 2&K5 K5 largesale
7 7:; 7:; = 7:;
M>
7P 7P 7 3 , , , 3 78
7
+  7: + 7:  5 5 + = , ), < 2&5, K5,  <5, , 7: 9 < 2&K5, K5, smal scale
; ;  ;
L O
FGH N>

Energy transfer from mean field to fluctuations

요 Positive Negative

Dissipation (Q) is always positive ⇒ sink of kinetic energy to thermal E

Molecular interaction

Energy flow
Geometry => Large-scales
Geometry - Mean KE - Production - Turb. KE - Dissipation (Q) - Thermal E
A hint of "energy cascade"
 Energy cascade
Behind an airfoil Source: Van Dyke, Album of Fluid Motions

From “ Turbulent flows” by Pope

Diagram of energy cascade

largescale
smalscale From “ Turbulent flows” by Pope
 Energy cascade
Gas turbine combustion

Source: J. Kim, Class material for "Introduction to Turbulent Flows"

 Energy cascade
Turbulent wake behind a walking human
From the scales of height and width to small-scales
 Energy cascade
Mixing of coffee and milk

dissipation
range
energy-containing
range

Source: J. Kim, Class material for "Introduction to Turbulent Flows"

 Description of energy cascade
Turbulence can be treated to be composed of eddies of different sizes
Eddies of size  have a characteristic velocity scale R and time scale
SR (=/R )
For details...

Consider a problem with geometric scale ℒ and reference velocity V

Turbulent KE produced at largest scales

W ~ℒ and W ~V
Viscosity is not important

Large eddies are unstable and easy to break up, transferring their energy
to smaller eddies 나나 d
(
Dissipation at smallest scales
INEnergy
balance 하르냐 >
8X R
Energy transfer from large to successively smaller scales (until % 1)
Y
ㆆ ㅡ

R  / & % 1 ~ Inertia / Viscosity ~ E. cascade / Dissipation % 1

…

~ ~
 Description of energy cascade
Energy flow
Geometry => Large-scales
(1) Geometry → Mean KE → (2) Production → Turb. KE → (3) Energy
transfer from large to smaller scales → (At the smallest scales) (4)
Dissipation → (5) Thermal E
ㆆ
At smallest scales : R  / & % 1 ~ TKE from cascade / Dissipation % 1
과 …

②
5
ThermalE 3
n ⑩③
Q
ㅡ
 Scales in wall-bounded flows
Canonical flows
Turbulent boundary layer . singlewall
프
Turbulent channel flow couble wall
wall

wall

More complex flows

Aerondynamic vehicles
Marine locomotions
Turbine, compressor
Pipe, pipeline, straw
River

Courtesy of S. Hahn, SNU

Summary: Basics rules on grid for accuracy
Rule #1. The smaller, the more accurate
Rule #2. Based on the size of spatial derivative
Rule #3. Based on the scale in space
Example: A flow around a circular cylinder
Contours of the x-velocity

Necessary grid spacing?

Length scale in x : similar to the distance between red spots (high U)

– About 5 (m)
If we want more than 20 intervals in a length scale
– Δx <= 0.25 (m)
 Near-wall scales in wall-bounded flows
Near the wall, viscous term is never small at a very high Re
…
3 7 ? 89 3 8
Viscous term ≅ ㅇ [\ 7] ?
~ [\ ^?
→

1/Re ↓ & velocity scale ↓ (wall : no-slip)

Length scale  gets smaller to balance them
Smallest scale near the wall

From energy cascade

_` 8a
Smallest scale : Inertial-viscous balance Y ≡ 1
Y
fY & d ~ Smallest length & velocity scales
8
At the wall : Sb ≅ c _a
`

de
Friction velocity : d ≡ =

Y =
Viscous length scale : fY  8  & ⇒Useful to determine wall grid spacing
a de
 Applications in wall-bounded flows
Grid size condition for near-wall grid resolution
fY ~ Smallest (Kolmogorov) length scale
∆] 8a ∆]
∆u N  _`
 Y
v1
Universal

Related non-dimensional variables : wall-units

] 8a ]
Length in wall unit : u N  _ 
` Y
8
Velocity in wall unit : N  8
a

Smallest (Kolmogorov) length scale away from wall

In fully-developed wall-bounded flows
Kolmogorov length scale w N % xu N 3/y for u N z 50 (semi-empirical)
x : 0.41
Coherent structures in wall-bounded flows
Turbulent channel flow : streamwise and spanwise velocities

Most energetic (energy-containing) near-wall structures: streamwise vortices

Elongated structures (in ) of counter-rotating vortices

| N % 30
} N % 100
N % 1000

Courtesy of D. Lee in SNU

Dimensionless small-scale descriptions are useful in terms of "Universality"

Very large-scale motions in wall-bounded flows
Jimenez (1998), Kim & Adrian (1999), etc.
Streamwise vortices : Diameter ~ Δ N % 30
Main turbulent motions : v 0.1f
Very large-scale motions (VLSM) : z f
In pipe, channel & boundary layer, VLSM carry
40–65% of the kinetic energy
30–50% of Reynolds shear stress
Visualization by Lee et al. (2013)