Wall Effects in Cavity Flows and their Correction Rules

The wall effects in cavity flows have been long recognized to be more important and more difficult to determine than those in single-phase, nonseparated flows. Earlier theoretical investigations of this problem have been limited largely to simple body forms in plane flows, based on some commonly used cavity-flow models, such as the Riabouchinsky, the reentrant jet, or the linearized flow model, to represent a finite cavity. Although not meant to be exhaustive, references may be made to Cisotti (1922), Birkhoff, Plesset and Simmons (1950, 1952), Gurevich (1953), Cohen et al. (1957, 1958), and Fabula (1964). The wall effects in axisymmetric flows with a finite cavity has been evaluated numerically by Brennen (1969) for a disk and a sphere. Some intricate features of the wall effects have been noted in experimental studies by Morgan (1966) and Dobay (1967). Also, an empirical method for correcting the wall effect has been proposed by Meijer (1967). The presence of lateral flow boundaries in a closed water tunnel introduces the following physical effects: (i) First, in dealing with the part of irrotational flow outside the viscous region, these flow boundaries will impose a condition on the flow direction at the rigid tunnel walls. This "streamline-blocking" effect will produce extraneous forces and modifications of cavity shape. (ii) The boundary layer built up at the tunnel walls may effectively reduce the tunnel cross-sectional area, and generate a longitudinal pressure gradient in the working section, giving rise to an additional drag force known as the "horizontal buoyancy." (iii) The lateral constraint of tunnel walls results in a higher velocity outside the boundary layer, and hence a greater skin friction at the wetted body surface. (iv) The lateral constraint also affects the spreading of the viscous wake behind the cavity, an effect known as the "wake-blocking." (v) It may modify the location of the "smooth detachment" of cavity boundary from a continuously curved body. In the present paper, the aforementioned effect (i) will be investigated for the pure-drag flows so that this primary effect can be clarified first. Two cavity flow models, namely, the Riabouchinsky and the open-wake (the latter has been attributed, independently, to Joukowsky, Roshko, and Eppler) models, are adopted for detailed examination. The asymptotic representations of these theoretical solutions, with the wall effect treated as a small correction to the unbounded-flow limit, have yielded two different wall-correction rules, both of which can be applied very effectively in practice. It is of interest to note that the most critical range for comparison of these results lies in the case when the cavitating body is slender, rather than blunt ones, and when the cavity is short, instead of very long ones in the nearly choked-flow state. Only in this critical range do these flow models deviate significantly from each other, thereby permitting a refined differentiation and a critical examination of the accuracy of these flow models in representing physical flows. A series of experiments carefully planned for this purpose has provided conclusive evidences, which seem to be beyond possible experimental uncertainties, that the Riabouchinsky model gives a very satisfactory agreement with the experimental results, and is superior to other models, even in the most critical range when the wall effects are especially significant and the differences between these theoretical flow models become noticeably large. These outstanding features are effectively demonstrated by the relatively simple case of a symmetric wedge held in a non-lifting flow within a closed tunnel, which we discuss in the sequel