22.10.2014 Views

The significance of coherent flow structures for the turbulent mixing ...

The significance of coherent flow structures for the turbulent mixing ...

The significance of coherent flow structures for the turbulent mixing ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

"<br />

Æ<br />

Û<br />

;<br />

;<br />

_^<br />

;<br />

§<br />

õ<br />

;<br />

(<br />

9<br />

9<br />

A<br />

"<br />

§<br />

"<br />

§<br />

§<br />

;<br />

§<br />

;<br />

(<br />

:<br />

:<br />

"<br />

Ø<br />

;<br />

;<br />

Æ<br />

÷<br />

;<br />

õ<br />

?<br />

§<br />

><br />

9<br />

§<br />

;<br />

§<br />

§<br />

:<br />

a`<br />

"<br />

<br />

A<br />

é<br />

Û<br />

;<br />

;<br />

§<br />

;<br />

;<br />

(<br />

"<br />

"<br />

é<br />

A<br />

;<br />

;<br />

õ<br />

9<br />

õ<br />

:<br />

<br />

;<br />

õ<br />

§<br />

Ê<br />

;<br />

õ<br />

:<br />

"<br />

5 Investigation <strong>of</strong> <strong>the</strong> xy-plane<br />

(5.17), into <strong>turbulent</strong> fluctuations is described by <strong>the</strong> following equation, which follows from<br />

equation (5.8) after <strong>the</strong> multiplication with õ ,<br />

õ Ê<br />

õ Ê<br />

õ ;<br />

6 ;<br />

Çõ> ?<br />

Ê ( Ê)C ;<br />

( ;<br />

õ]Ê<br />

ced f<br />

b<br />

loss <strong>of</strong> kinetic energy <strong>of</strong> <strong>the</strong> mean motion<br />

cd f<br />

b<br />

production <strong>of</strong> turbulence<br />

Ê õ Ê<br />

ced f<br />

b<br />

transfer <strong>of</strong> kinetic energy<br />

cd f b<br />

direct dissipation<br />

Ê <br />

9hg<br />

( Ê CZi<br />

Ø ¾ (5.18)<br />

( C <br />

é í ;<br />

:jA<br />

: Ê<br />

:)ï<br />

and <strong>the</strong> generation, transport and decay ( <strong>of</strong> can be derived from <strong>the</strong> Navier-Stokes equation<br />

when <strong>the</strong> equation <strong>for</strong> <strong>the</strong><br />

component multiplied by be<strong>for</strong>e averaging.<br />

component is multiplied by ( and added to <strong>the</strong> equation <strong>for</strong> <strong>the</strong> (<br />

(5.19)<br />

õ ;<br />

6 ;<br />

( Ê ;<br />

cd f<br />

b<br />

convection<br />

cd f<br />

b<br />

redistribution<br />

ced f b<br />

production<br />

ö†í ;<br />

9 ï<br />

§1" §<br />

" §<br />

A b ced f :<br />

diffusion<br />

ced f<br />

b<br />

dissipation<br />

k Ê ( C<br />

(4k Ê<br />

;<br />

( Ê<br />

ökà[÷ C<br />

Un<strong>for</strong>tunately, <strong>the</strong> remaining ma<strong>the</strong>matical difficulty, associated with <strong>the</strong> statistical approach,<br />

is that <strong>the</strong> dynamical equations are under-determined due to <strong>the</strong> non-linearity in <strong>the</strong> momentum<br />

equations. In case <strong>of</strong> equation (5.8), <strong>for</strong> example, a relation between <strong>the</strong> õ<br />

mean §QDF§lG motion<br />

and all dominant components <strong>of</strong> <strong>the</strong> general Reynolds-stress tensor have to be assumed<br />

be<strong>for</strong>e a numerical integration can be per<strong>for</strong>med. From <strong>the</strong> present point <strong>of</strong> view it seems<br />

unlikely that this so called closure problem can be solved from first principles, because most<br />

<strong>of</strong> <strong>the</strong> developed ma<strong>the</strong>matical approximation methods fail <strong>for</strong> two reasons. Firstly, <strong>turbulent</strong><br />

shear <strong>flow</strong>s appear as a non-homogeneous, non-isotropic, multi-scale problem without any real<br />

scaling similarity ranking from <strong>the</strong> largest to <strong>the</strong> smallest scales. Secondly, <strong>turbulent</strong> <strong>flow</strong>s are<br />

generally global and not local, which means that <strong>the</strong> local properties are not governed by <strong>the</strong><br />

mean motion. For <strong>the</strong>se reasons, it is evident that <strong>the</strong> functional dependence and §QDF§lG<br />

statistical<br />

structure <strong>of</strong> must be determined experimentally. This will be <strong>the</strong> prime issue in <strong>the</strong><br />

following.<br />

5.2 Experimental set-up<br />

<strong>The</strong> principle set-up <strong>of</strong> <strong>the</strong> double stereo-scopic PIV system utilised <strong>for</strong> <strong>the</strong> investigation in <strong>the</strong><br />

wind-tunnel displayed on page 8, is shown in figure 5.1. Because <strong>of</strong> <strong>the</strong> size <strong>of</strong> <strong>the</strong> large-scale<br />

<strong>structures</strong> in stream-wise direction and <strong>the</strong> relatively small velocity variations which must be<br />

properly resolved, <strong>the</strong> system is installed side by side. <strong>The</strong> actual observation â angles and<br />

distances along with <strong>the</strong> exact coordinates <strong>of</strong> each camera sensor, relative to <strong>the</strong> centre<br />

?nm<br />

<strong>of</strong> <strong>the</strong> corresponding observation area, are displayed in table 5.2. For <strong>the</strong> registration and<br />

storage <strong>of</strong> <strong>the</strong> particle images four Peltier cooled high resolution PCO cameras with 1280<br />

72

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!