14.01.2015 Views

LES of shock wave / turbulent boundary layer interaction

LES of shock wave / turbulent boundary layer interaction

LES of shock wave / turbulent boundary layer interaction

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 3<br />

Flat plate <strong>boundary</strong> <strong>layer</strong><br />

A well known problem in the simulation <strong>of</strong> <strong>turbulent</strong> <strong>boundary</strong> <strong>layer</strong>s<br />

is that realistic inflow data are needed. Recently Xu & Martin (2004)<br />

have reviewed different inflow generation techniques. Following them,<br />

the inflow generation techniques can be organized into three categories:<br />

1 spatially evolving <strong>boundary</strong> <strong>layer</strong> simulating the full transition process;<br />

2 using data from previous simulations (or data combined with those<br />

from experiment);<br />

3 various techniques using an instantaneous downstream flow field<br />

(streamwise periodic <strong>boundary</strong> conditions, extended temporal approach,<br />

rescaling and recycling method, etc.).<br />

A brief summary <strong>of</strong> available simulations can be found in appendix A.<br />

A DNS <strong>of</strong> a spatially evolving <strong>turbulent</strong> supersonic flat-plate <strong>boundary</strong><br />

<strong>layer</strong> is conducted by Rai et al. (1995). A developed <strong>turbulent</strong> <strong>boundary</strong><br />

<strong>layer</strong> with a momentum thickness <strong>of</strong> Re θ ≈ 4500 was obtained by the<br />

simulation <strong>of</strong> laminar-<strong>turbulent</strong> transition initiated by a blowing/suction<br />

strip. This case is close to an experiment by Shutts et al. (case 55010501<br />

in Fernholz & Finley (1977)) with Re θ ≈ 6000 and M ∞ = 2.25. Later<br />

Gatski & Erlebacher (2002) and Pirozzoli et al. (2004) repeated this DNS<br />

on a finer grid. Their results showed that both the near-wall asymptotic<br />

behavior and the log-law exhibit similarities with the incompressible case<br />

when the van-Driest velocity transformation is applied. Similarly, the<br />

Reynolds stresses also are independent <strong>of</strong> Mach number when scaled with<br />

the mean density ratio as suggested by Morkovin (1962). The results also<br />

showed that the total temperature fluctuations were not negligible. <strong>LES</strong><br />

<strong>of</strong> this case was performed by Spyropoulos & Blaisdell (1998) using a<br />

dynamic Smagorinsky model and by Rizzetta & Visbal (2004), using<br />

Smagorinsky, dynamic Smagorinsky and MI<strong>LES</strong> approaches.<br />

Guarini et al. (2000) employed a ”slow growth” assumption to simulate<br />

a <strong>turbulent</strong> <strong>boundary</strong> <strong>layer</strong> at Re θ = 1577. It is assumed that the<br />

<strong>boundary</strong> <strong>layer</strong> grows slowly in the streamwise direction so that the turbulence<br />

can be treated as approximately homogeneous in this direction.

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

Saved successfully!

Ooh no, something went wrong!