05.11.2012 Views

Synthetic Inflow Condition for Large Eddy Simulation (Synthetic - KTH

Synthetic Inflow Condition for Large Eddy Simulation (Synthetic - KTH

Synthetic Inflow Condition for Large Eddy Simulation (Synthetic - KTH

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.

CHAPTER 2. BACKGROUND<br />

turbulence models used to close the equations are provided. Although the software<br />

elsA solves the compressible Navier-Stokes equation, we will restrict ourselves to<br />

the simulation of incompressible flows where the mass density ρ and the dynamic<br />

viscosity µ are constant. In this case, the Navier-Stokes equations governing the<br />

evolution of the velocity u and pressure p of the fluid read<br />

ρ ∂u<br />

∂t<br />

+ ∇.(ρu ⊗ u) = −∇p + ∇.σ (2.1)<br />

∇.u = 0 (2.2)<br />

The viscous stress tensor is σ = 2µS <strong>for</strong> a Newtonian fluid where<br />

is the strain rate tensor.<br />

S = 1<br />

2 (∇u + (∇u)T ) (2.3)<br />

However since the direct numerical simulation of the above Navier-Stokes equations<br />

requires an extreme computational cost due to the several scales involved, a<br />

smoothing operator has to be applied to the exact solution u of the Navier-Stokes<br />

equations. The complexity of the system to be solved will thus be reduced.<br />

Assuming that the LES spatial filter and the derivative operators commute,<br />

governing equations <strong>for</strong> the filtered/averaged quantities u and p can then be derived,<br />

where u corresponds to the LES averaging operators<br />

ρ ∂u<br />

∂t<br />

+ ∇.(ρu ⊗ u) = −∇p + ∇.(σ + τ) (2.4)<br />

∇.u = 0 (2.5)<br />

And the subgrid-scale stress tensor is τ = −ρ(u ⊗ u − u ⊗ u)<br />

Hence the LES representations of the flow field are seen as two levels of description<br />

of the exact solution u of the Navier-Stokes equations.<br />

Labourasse and Sagaut [20] defined the LES spatial filtering operator (see Eq.<br />

(2.6)) in the general framework of multilevel methods relying on different scale<br />

separation operators.<br />

In fact the separation of large scales and small scales is done via a low-pass<br />

filtering operation, which is defined as a convolution product<br />

� +∞<br />

u = G ⋆ u = u(y)G△ (x − y)dy (2.6)<br />

−∞<br />

4

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

Saved successfully!

Ooh no, something went wrong!