16.08.2013 Views

The UMIST-N Near-Wall Treatment Applied to Periodic Channel Flow

The UMIST-N Near-Wall Treatment Applied to Periodic Channel Flow

The UMIST-N Near-Wall Treatment Applied to Periodic Channel Flow

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

CHAPTER 5. RESULTS 69<br />

<strong>The</strong> solution procedure employed involved initialising flow variables <strong>to</strong> zero,<br />

applying the driving pressure gradient, and iteratively updating the solution<br />

until convergence was achieved. Gradients in time ( ∂<br />

∂t<br />

terms) were included<br />

in the solution, so that unconverged results may have been regarded con-<br />

ceptually as the transient response of the flow <strong>to</strong> a step change in pressure.<br />

However, transient results are not shown. ∂<br />

∂t<br />

terms do not affect the con-<br />

verged result in steady flow and were included primarily for debugging.<br />

In Figure 5.1, the k-ε solutions overpredict peak velocity and bulk flow, while<br />

the k-ω model underpredicts. Figure 5.3 shows that the 〈U〉 + results nearer<br />

<strong>to</strong> the wall are similar for each model and match the DNS result closely.<br />

Equation 3.17 can be evaluated at y = 0 <strong>to</strong> obtain<br />

<br />

d 〈U〉 y=0<br />

τw = ρν<br />

dy<br />

d〈P 〉<br />

dx by<br />

Recalling Equation 3.19, τw is related <strong>to</strong><br />

<br />

d 〈P 〉<br />

τw = −δ<br />

dx<br />

(5.1)<br />

(5.2)<br />

<strong>The</strong>refore, close agreement on 〈U〉 + near the wall is <strong>to</strong> be expected from the<br />

prescription of<br />

d〈P 〉<br />

dx . <strong>The</strong> deviation in 〈U〉+ occurs farther from the wall and<br />

is likely due <strong>to</strong> the prediction of νt.<br />

<strong>The</strong> subgrid solution produces results that are very similar <strong>to</strong> those of the<br />

low-Reynolds-number approach. This is <strong>to</strong> be expected, since the internal<br />

subgrid calculation is identical <strong>to</strong> a standard low-Reynolds-number calcu-<br />

lation in a parabolic solution scheme. <strong>The</strong> unique aspect of the subgrid<br />

approach as compared <strong>to</strong> a low-Reynolds-number parabolic solution is in the<br />

transferral of information between the subgrid and the main grid via mutu-<br />

ally dependent boundary conditions. However, this complexity is unlikely <strong>to</strong>

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

Saved successfully!

Ooh no, something went wrong!