order to extrapolate the results to full scale? In many papers, authors indicate that oxygen content acts on cavitation inception ([2], [3], [6], [9]) and on desinence cavitation[13]. We could ask if there is a best water quality for model tests. But the above mentionned studies do not permit to quantitatively correlate a parameter describing water quality and a cavitation number.

Many studies were conducted on different French tests facilities during the Action Concertée Cavitation. The aim was a better understanding of tip vortex roll-up and of associated cavitating phenomenon ([ 7], [8], [11]). Due to the fact that the water tunnel, the Grand Tunnel Hydrodynamique G.T.H., permits us to control the oxygen content independently from the freestream nuclei content, the Bassin d'Essais des Carènes was in charge of studing relation between oxygen content, freestream nuclei content and inception, development and desinence of cavitation [11].

The aim of this paper is first to recall the main results obtained during these tests. For the cavity within the tip vortex-roll up, a law was yet proposed as a function of the Reynolds number for a single water quality at the desinence cavitation [8]. We will see in this paper, how all the data for cavitation inception can be fitted on a single curve when taking into account freestream nuclei.

The diameter of the cavity within the vortex was also measured. This allowed to appreciate how the water quality affects the development of cavitation. Thus, the hysteresis at desinence can be explain by diffusion of non condensable gas as expected by Holl [13].

At the inception of the cavity, we observe bubbles growing in the center of the tip vortex for special water quality. These bubbles, captured by the vortex, could grow by diffusion of non condensible gas or could explose by vaporisation of liquid. We are particulary interested in the acoustic noise and, in the following, we will call cavitation inception, a phenomenon with two particularities: we can locate the tip vortex in the flow visually, and there is a raise of the acoustic noise.

For developed cavitation, specific frequencies were extracted in the spectrum signal. These frequencies are compared with the frequencies of modal deformations of the cavity.

Experiments were conducted in the G.T.H. [5], in the square test section of 1.14 m width, 1.14 m high and of 6 m long. The studies were made with an elliptical planform hydrofoil, the maximum chord is 475 mm and the half-span equals 753 mm (the aspect ratio is 3.8), the cross-section is a NACA 0020.

The tunnel permits us to control the inlet velocity between 2 m/s up to 20 m/s with an accuracy of 0.2%. The pressure level varies between 2.10⁴ Pa up to 4.5 10⁵ Pa, with an accuracy of 0.5%. The cavitation number σ is defined as:

(1)

with P, the total inlet pressure in the plane of the tip of the hydrofoil, Pv, the vapor pressure of water, ρ the volumic mass of water and V the inlet flow velocity. The precision of σ is 1%.

The oxygen content is controlled between 25% up to 160% of saturation and is measured with an oxygen probe. The flow can be seeded with nuclei through injectors located upstream the tunnel nozzle. A Cavitation Susceptibility Meter based on a centerbody venturi (called VAG) measures the nuclei distribution in the fluid. Hence, the number of nuclei with a given critical pressure is known for each test configuration. Several nuclei injections patterns were tested. They are characterised by a specific susceptibility pressure, a concentration and a distribution in the test section. Hereafter, the higher Ps (lower tension) corresponds to the water with ‘big' bubbles, then we have a medium Ps for ‘small' bubbles and the water without injection have the lowest Ps (higher tension). For more details, see [3].

Cavitating data were collected with the two testing configurations of nuclei presented below.

The hydrofoil is mounted vertically at the top of the test section (Figure 1). The angle of attack is changed