Problèmes de perte de charge et de stabilité des grilles de prise d'eau

Water intake screen head loss and stability problems

Chargé de cours à l'ecole Nationale Supérieure des Mines de Saint-Etienne ; Chef du Département Recherches de la Société B.V.S.

Résumé

Definitions of the various screen (trash rack) elements are reviewed, also the eight parameters governing loss of head across a screen. Five of these parameters are "solid" area s, "open" area b, length L (see Fig. 2), vertical angle e (Fig. 1) and bar cross-section (Fig. 3), which Thoma and Kirschmer [2] took into account at Munich Institute in 1926 The sixth parameter, horizontal angle θ (Figs. 1 and 4) was studied by Thoma and Spangler in 1928 [3]. Measurements in Situ gave very much higher head losses than in the laboratory, and the explanation for this was provided by Dulnyev [6] and Berezinski [4] and [7], namely the failure to allow for the considerable amount of obstruction due to horizontal screen stiffeners, bracing members, fittings, etc. The seventh parameter, p, is the "solid", percentage of the total area, which varies between 20 and 40 % whereas the percentage accounted for solely by upright bars only makes up about 6 to 16 % of the total. The eighth parameter is the "residual detritus" factor Kd which indirectly expresses cleaning efficiency. The loss of head across a screen perpendicular to the flow is given reasonably well by formula (1), in which : Kd = 1.1 to 1.2, for a modern automatic screen cleaner (trash rack rake) ; Kd = 1.5, for older designs ; Kf = 0.51, for a "long" rectangular cross-section ; Kf = 0.35, for a circular cross-section ; Kf = 0.32, for an clongated cross-section with semi-circular ends. The function f (L/b) is shown in Figure 2. A numerical example is worked out for an operational installation at the end of the article. Figure 3 gives first approximates for the head loss coefficient [9]. Figure 4 gives data required to calculate the above coefficient for oblique inflows to the screen, with an example of its application ([3], [6] and [11]). Screen stability, which was frequently at fault before 1955, is discussed at length. The author has shown [1] this to be due to resonance of the natural bar frequency and the frequency of alternating eddy wakes forming on the fear face of the bar. The natural frequency of long components in a vacuum (actually in air) was given by Timoshenko in 1950. The author's formula (5) allows for the fact that such bars vibrate in water [13]. The meanings of symbols in this formula are as follows r = radius of gyration of the section ; H = distance between bracing members ; E and Yb = elastic modulus and density of bar material ; y' = fluid density (0.001 kg/cm3) for water) ; M = "Fixing factor" resulting from (6) and (7) for fixed and articulated ends respectively. Figure 5 [13] gives f in cycles/sec. for b/s = 10, and f is found for other values in this report by a simple calculation, which is demonstrated in the final numerical example. Frequency is very much lower in water than in air (by a factor varying between 0.6 and 0.9). Formula (5) applies for b/L ≤ 0.7, but above this value the calculation is carried out with (b/L) = 0.7.