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 79<br />

<strong>The</strong> log law offers predictions of k + in Figures 5.23 & 5.24 that are reasonable.<br />

<strong>The</strong> log law solution has also been improved greatly by the prescription of U<br />

rather than<br />

∂〈P 〉<br />

. It must once again be noted that the log law only produces<br />

∂x<br />

interesting results away from y/δ = 0.2, where it is applied for this particular<br />

flow.<br />

A ‘kinked’ region of large change in gradient tends <strong>to</strong> exist in many of the<br />

profiles of 〈U〉 + . This phenomenon can be seen most clearly in Figures 5.25<br />

& 5.26. It is produced by all of the models except for the log law treatment,<br />

and is particularly prevalent in the profiles produced by the low-Reynolds-<br />

number k-ε model. A kink indicates an underprediction of the diffusion of<br />

〈U〉 + within the vicinity of the kink. Deferring <strong>to</strong> the assumptions of the<br />

EVM, this would suggest a local underprediction of k + . This corollary can<br />

in fact be seen in the figures. Comparing phase angles of 3π/4, π, and 5π/4,<br />

suggests that the location of a kink in 〈U〉 + corresponds <strong>to</strong> a location of<br />

maximum underprediction of k + . This may be seen by examining Figures<br />

5.25 & 5.26 for velocity and 5.27 & 5.28 for k + .<br />

<strong>The</strong> same phenomenon of a kinked velocity profile can be observed indirectly<br />

in Figures 5.21, 5.22, 5.23 & 5.24, where 〈U〉 + predicted by the low-Reynolds-<br />

number k-ε model appears <strong>to</strong> lead the DNS result in phase closer <strong>to</strong> the wall,<br />

whereas it lags the DNS in phase at higher y/δ. This apparent peculiarity is<br />

in fact a symp<strong>to</strong>m of the changing slope of the 〈U〉 + profile produced by the<br />

k-ε model at different phase angles.<br />

All of the models produce flatter velocity profiles than the DNS, as seen in<br />

Figures 5.25, 5.26, 5.29 & 5.30. This is particularly true just after a phase<br />

angle of π, when the flow is subjected <strong>to</strong> an adverse pressure gradient and a

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

Saved successfully!

Ooh no, something went wrong!