Quelques nouveautés sur les déversoirs pour la mesure des débits

Directeur du Laboratoire de Banlève de l'Institut de Mécanique des Fluides de Toulouse.

Résumé

I-The two most popular sharp-crested weirs in use today are the following : (i) Rectangular weirs with or without end contractions; (ii) 90 degree Vee-notches. The empirical formulae for these weirs give the discharge in terms of weir geometry and the depth of water on the sill. These formulae, which are the outcome of much experimental work by numerous researchers, include a discharge coefficient depending on h and p and implicitly allowing for the velocity head Vo2/2 g in the cross-section in which h is measured. Formulae 1, 2, 3 and 4 were first proposed by various authorities ; they apply to weirs without end contractions, whilst formulae 5 and 6 are for weirs with end contractions. Table 1 compares the calculated discharge data for various values of h and p and shows that there may be as much as 2 % error. These formulae are of limited applicability and, in particular, cannot he used with high approach velocities in the channel. II-Recent experimental work at the Toulouse Fluid Mechanics Institute has extended the possible measurement range by application of a relationship (formula 7) allowing explicitly for the upstream approach velocity term V2/2 g. This relationship was checked on various test installations and the mean error was found to be less than 1 per cent. Experiments with Vee-notches have been run at the Ecole Nationale Supérieure Agronomique at Rennes. A theoretical study based on assumed velocity distributions at the weir crest produced formula 8, the numerical coefficients of which were corrected from the experimental data. Formula 9 allows explicitly for V2/2 g. It enables measurement even for p = O. Discrepancies between experimental discharge data and as calculated by formula 9 are between - 1.4 % and + 0.5 % for the considered experimental conditions. Formulae 8 and 9 enable the use of sharp-crested rectangular weirs or Vee-notches over a much wider measurement range than the conventional formulae now in use. When these various formulae were compared, there was found to he a 2 % measurement error, which can only be reduced by calibration of each weir structure before use.