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

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

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

CHAPTER 5. RESULTS 81<br />

maxima). This is particularly visible in Figure 5.31, but Figure 5.27 shows<br />

that the local minimum between these two peaks lies at y/δ = 0.2, the limit<br />

of the subgrid region and also the point where boundary conditions from the<br />

subgrid are applied <strong>to</strong> the main grid and visa versa. Close inspection of all<br />

the k + profiles produced by the k-ω subgrid reveals minor discontinuities at<br />

this y/δ location for other phase angles also. This is a result of the way<br />

in which the subgrid solution of ω was applied as a boundary condition <strong>to</strong><br />

the main grid. <strong>The</strong> <strong>UMIST</strong>-N approach <strong>to</strong> k and ε boundary conditions<br />

is <strong>to</strong> calculate subgrid averaged production and dissipation terms for these<br />

parameters, which are then applied as source terms within the near-wall cell<br />

in the main grid solution. However, ω tends <strong>to</strong> infinity at the wall, and<br />

source terms on ω become numerically unwieldy in near-wall cells. Thus,<br />

subgrid averaged production of ω could not be obtained and instead the<br />

subgrid profile of ω was interpolated <strong>to</strong> the main grid node and applied there<br />

as a fixed value. 2 This boundary condition appears <strong>to</strong> be less effective than<br />

the average source approach, and, until a more satisfac<strong>to</strong>ry solution <strong>to</strong> the ω<br />

boundary condition problem can be found, it advises against the applicability<br />

of the <strong>UMIST</strong>-N approach <strong>to</strong> a k-ω model solution. This aside, the subgrid<br />

k-ω results appear satisfac<strong>to</strong>ry when considered as a whole.<br />

In addition <strong>to</strong> the above difficulty in the prescription of an ω boundary<br />

condition, further difficulties were encountered in attempting <strong>to</strong> run the k-ω<br />

model in the subgrid with a high-Reynolds-number k-ε model in the main<br />

grid. Poor results were seen <strong>to</strong> arise from the difference in the values of k<br />

produced by the two models under similar conditions. In general, the k-<br />

ω model produces lower values of k than the k-ε model, as noted in the<br />

2 See Chapter 4 for more information.

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

Saved successfully!

Ooh no, something went wrong!