pdf, 12 MiB - Infoscience - EPFL
pdf, 12 MiB - Infoscience - EPFL
pdf, 12 MiB - Infoscience - EPFL
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Summary and conclusions<br />
scour, was also enhanced. The correlation for the maximum scour depth can be increased to 0.82<br />
and the predicted cross-section shape fits quite well to the measured one (without macro-roughness).<br />
A new way was explored to establish a formula based on the shape of the cross-section (§ 7.3.3).<br />
Assuming that the maximum bed slope in a radial direction is limited to a maximum value corresponding<br />
to the friction angle of the bed material, different polynomial functions (Table 7.6) of<br />
the third and fifth degree were tested. The best correlation ( R 2 = 0.856 )was obtained with a<br />
polynome of the third degree (equation 7.31), fitting well to the measured cross-section. But the<br />
application to results of scale model tests and field measurements showed a significant underestimation<br />
of the maximum scour depth. Therefore the modified formula of Bridge (eq. 7.7) is proposed<br />
for the scour depth computation for the configuration without macro-roughness. This<br />
equation showed a good agreement with field data (§ 7.5)<br />
sinβ<br />
0.394 11 23 h m<br />
= ⋅ ⎛ – ⋅ ----- ⎞ -----<br />
⎝<br />
, (7.7)<br />
B ⎠<br />
⋅ ⋅ tanφ<br />
⋅ ---- R<br />
B r<br />
2 = 0.817<br />
Another approach, inspired by the similitude and approximation theory of KLINE (1965) was<br />
investigated (§ 7.3.4), resulting in physically based dimensionless parameters. Unfortunately the<br />
resulting correlations are quite low.<br />
Finally a genetic algorithm written by KEIJZER & BABOVIC (1999) was used to search for a function<br />
fitting well to the measured data (§ 7.3.5). Without ribs, the previously obtained results could<br />
not be enhanced. The search for an equation predicting the scour locations was more successful,<br />
resulting in two formulae with good correlations.<br />
Finally a formula was established, allowing for the prediction of the maximum scour depth in the<br />
presence of macro-roughness with an excellent result ( R 2 = 0.876 ):<br />
R c<br />
h s<br />
h<br />
---------- max<br />
7.7 e s<br />
= ⋅ ----- ⋅ Fr ⋅( 0.001 + ( θ–<br />
θ<br />
h m<br />
R cr ) 2 ) + 1.7<br />
h<br />
(7.63)<br />
Enhancing the results obtained with the genetic algorithm, the following equations were established<br />
for the determination of the location of the scour holes without macro-roughness:<br />
B S<br />
α 1 σ 0.58 -----<br />
e<br />
= ⋅ ⎛ ⋅ + <strong>12</strong>.7 ⋅ ---------------- ⎞<br />
⎝<br />
, (7.54)<br />
h m σ – Fr d<br />
⎠<br />
+ 1.4 ⋅Fr V<br />
d ⋅ ------ – 6.6 R 2 = 0.829<br />
V∗<br />
R c<br />
α 2 <strong>12</strong>.6 ⋅ Fr d 0.9 ⎛-----<br />
⎞ 2<br />
= – ⋅ + 91.6 , R (7.57)<br />
⎝ B ⎠<br />
2 = 0.602<br />
For the second scour, a slightly better correlation can be obtained with equation (7.56) but which<br />
is much more complicated and quite sensitive to the choice of the used parameters.<br />
<strong>EPFL</strong> Ph.D thesis 2632 - Daniel S. Hersberger November 9, 2002 / page 197