21.11.2013 Views

pdf, 12 MiB - Infoscience - EPFL

pdf, 12 MiB - Infoscience - EPFL

pdf, 12 MiB - Infoscience - EPFL

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 7 - Establishing an empirical formula<br />

The shape of the last equation gets quite close to the parabolic form of many traditional equations.<br />

To allow an appreciation of the quality of the obtained relation, an equation with corrections<br />

(eq. 7.20) is given on Fig. 7.7.<br />

The so obtained correlation for the maximum scour depth can be compared to the empirical formula<br />

of PETER (1986), who obtains a correlation of 0.87 compared to the complete data set.<br />

Hoping to obtain even higher correlations with other functions, the following two type of functions<br />

were explored: a polynomial function of the 3rd degree, but with additional terms<br />

( y = x 3 + x 2 + x ) and a polynomial function of the 5th degree.<br />

d) Polynomial function of the 3rd degree with additional terms - uncentered with vertical<br />

adjustment<br />

The following function was used with the boundary conditions mentioned in paragraph c).<br />

Solving for the different constants, leads to:<br />

h s = h m + tanφ∗<br />

⋅<br />

c 1<br />

c<br />

h s<br />

---- ⋅ ( r–<br />

c<br />

3 2<br />

) 3 3<br />

=<br />

+ ---- ( r–<br />

c<br />

2 2<br />

) 2 + c 4<br />

⋅ ( r–<br />

c 2<br />

) + c 5<br />

4<br />

--------<br />

3B 2<br />

1<br />

⋅ ----------------<br />

4ξ 2 ⋅ ( r – R<br />

– 1 c – ξB) 3 ---<br />

4 ξ<br />

+ ⋅ ----------------<br />

B 4ξ 2 ⋅ ( r– R<br />

– 1 c – ξB) 2<br />

B<br />

( r – R c – ξB)<br />

---<br />

4ξ 3 – 3ξ<br />

+ + ⋅ -------------------<br />

3 4ξ 2 – 1<br />

(7.21)<br />

(7.22)<br />

dh<br />

------- s<br />

= tanφ∗<br />

dr<br />

⋅<br />

4<br />

-----<br />

B 2<br />

1<br />

⋅ ----------------<br />

4ξ 2 ⋅ ( r – R<br />

– 1 c – ξB) 2 ---<br />

8 ξ<br />

+ ⋅ ----------------<br />

B 4ξ 2 ⋅( r– R<br />

– 1 c – ξB)<br />

+ 1<br />

(7.23)<br />

The final solution has quite a strange behavior as far as the choice of ξ is concerned. At ± 50 %<br />

the curvature (second derivative) has a discontinuity. Therefore this value had to be avoided.<br />

Without any correction factor the correlation is very low (below 0.50) and the order of magnitude<br />

of the predicted scour does not fit at all. The correlation (maximum scour) can be increased up to<br />

0.81 by introducing a correction factor c = 4.45 ⋅ ( 690 ⋅ S eall ,<br />

+ 5.8 ⋅ R c ⁄ B – 79 ⋅ h m ⁄ B)<br />

, but<br />

the shape of the predicted bed does not fit well to the measured one (Fig. 7.8). With other correction<br />

factors, the bed shape fits better, but with lower correlations up to 0.75.<br />

Rendering the function more complicated does not bring the desired effect of a better prediction<br />

of the maximum scour nor a better fit to the transversal bed topography.<br />

page 166 / November 9, 2002<br />

Wall roughness effects on flow and scouring

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

Saved successfully!

Ooh no, something went wrong!