Une méthode de mesure des débits au moyen de moulinets dans les pertuis d'entrée des usines hydro-électriques de basse chute

A method of measuring discharge with current meters in low-head power plant intakes

E.D.F., Centre de Recherches et d'Essais de Chatou.

Résumé

The difficulties experienced in the measurement of discharge absorbed by vertical-shaft Kaplan units are mainly due to their large intake sluice dimensions and sharp convergence over a short distance (Fig. 1), which frequently result in very oblique flows in the gauging section. In order to reduce the time the power unit is out of operation to a minimum, continuous integration measurements are first carried out at the 25 to 30 conditions required for the efficiency curve determination, by means of self-compounding current meters mounted on a frame (Fig. 2) free to move in the stoplog grooves at about 1% of the flow velocity, so as to cover the entire depth water in the intake sluice. Though these measurements are not completely reliable because of the very wide incidence angle and turbulent flow, they nevertheless certainly give an adequate relative discharge value for fairly similar conditions, and can then be calibrated to give an absolute value by carrying out a few detailed measurements in the velocity field. This is done by positioning the current meter frame at a number of different horizontals in succession and determining first the direction of the velocity at each point and then its magnitude from the known angular response relationship for swept-back current meter propellers. For this, two measurements Vα1 and Vα2 are made with the current meter heading successively in two directions at 45° to each other, between which the true velocity direction is assumed to lie. As the response relationship (Vα/V0) = f (α) is known (Fig. 3), Vα1/Vα2 only depends on the angle α1 between the velocity and one of the current meter headings. The true magnitude of the velocity is then given by a third measurement in the direction thus determined; certain corrections have to be applied to allow for the slight mechanical angular error remaining between the third current meter heading and the velocity, and for the slight effect of the frame on the flow, which requires a preliminary calibration to be carried out for the velocity direction with respect to the frame instead of to the current meter. All that then remains is to calculate the normal velocity component, to repeat this operation for a certain Humber of suitably distributed horizontals, and to integrate the resulting velocity field graphically in order to find the exact discharge. By then comparing this with the rough value obtained by continuous integration at the same conditions, a correction factor is found for neighboring conditions, which can be assumed to feature the same velocity distribution. A slip angle too large to be neglected can be measured independently by a vane aligning itself automatically with the vertical velocity plane, but only if a set of laboratory calibration curves depending on two angular parameters is available instead of a curve merely far (Vα/V0) = f (α). Means of controlling and ascertaining the current meter headings from the surface are required for this method, which are provided by mounting the current meters on a round tube which can be made to rotate in steps of exactly 5° within a quadrant by a wheel with machined teeth spaced at 1/10°. Numerous preliminary calibrations had been carried out at the "Institut de Mécanique des Fluides" at Toulouse and the "Bassin d'Essai des Carènes" test tank in Paris, in which the current meters were towed in still water, and it was then decided to check the angular response relationship thus obtained in the actual power plant intakes. Swept-back propellers were preferred to the self-compounding type for the velocity field determinations, as the accuracy of the latter seemed to be affected by flaw turbulence. The above discharge measurement method has been used at three major power stations: Vogelgrün and Marckolsheim on the Rhine and Logis Neuf on the Rhone. Its main disadvantage is that the actual measurements and initial sorting of the results arc both rather unwieldy. For instance, about four hours are required to coyer the velocity field for one given set of conditions thoroughly, during which turbine operation must remain completely stable. This drawback is offset, however, by the fact that this method is alone in offering a really analytical picture of the flow, which is its main advantage. The results obtained so far are considered to be most promising for three reasons: 1) The overall results are very coherent as regards both continuous integration and velocity field exploration; 2) the individual velocity directions and magnitudes and the flow distribution between the three intake sluices measured on site agree very closely with the results of preliminary tests carried out on an air flow model at the "Centre de Recherches et d'Essais" at Chatou (Fig. 4); 3) comparable efficiencies were measured on both prototype and model turbines, which were transposed on the same basis for all three considered power stations. This method of measurement can therefore reasonably be expected to provide fairly reliable information for the checking of industrial Kaplan turbine efficiencies and, above all, to enable more to be found out about the scale effects associated with these turbines and their models.