A previously unpublished theory for describing the internal flow in a gas centrifuge is presented. The theory is based on boundary-layer-type arguments on the side walls of the centrifuge with the additional approximation of neglecting radial diffusion of radial momentum. The effects of the top and bottom end caps are incorporated through Ekman-layer solutions. The results are presented in a form amenable to numerical calculations.

Some sample calculations are presented for the special case of a centrifuge with a linear temperature profile on the wall and the top and bottom of the centrifuge at the same temperature as the corresponding end of the side wall.

The spreading on a water–air interface of a thin liquid film is examined for the situation in which surface tension gradients drive the motion. A similarity solution is obtained numerically for unidirectional spreading when some general restrictions concerning the form of the liquid film constitutive relation is made. This solution gives the size of the film as a function of time and also the velocity and thickness distribution along the spreading film. Experiments are performed which show good agreement with the theory.

The shape of a vertical slender jet of fluid falling steadily under the force of gravity is studied. The problem is formulated as a nonlinear free boundary-value problem for the potential. Surface tension effects are neglected. The use of perturbation expansions results in a system of equations that can be solved by an efficient numerical procedure. Computations were made for jets issuing from orifices in various shapes including an ellipse, a rectangle, and an equilateral triangle. Computational results are presented illustrating the propagation of discontinuities and the formation of thin sheets of fluid.

Non-stationary MHD interaction of a horizontal magnetic field with a three-dimensional cellular convection is studied by means of computational methods and methods of mean field electrodynamics.

For a given magnetic field drop across the convective layer, the rate of magnetic flux penetration through this layer is characterized by two integral coefficients: the first one describing the topological pumping effect arises from the antisymmetric part of the α-effect, while the second coefficient accounts for the enhancement of the effective diffusion due to the convective motions. In the magnetic-Reynolds-number range studied (−5 [les ] Rm [les ] 5) these coefficients are found to be, correspondingly, odd and even functions of Rm only. The net magnetic flux escape rate into vacuum decreases at Rm > 2·2 when compared with a case of a layer without cellular motions. Here the topological pumping prevails not only over the convective enhancement of diffusion but begins to suppress even the background diffusion action.

Thus, the asymmetry in the transport properties of cellular motion is again demonstrated, and their difference from those of random turbulence is identified.

This paper describes a new balance, suitable for direct measurement of skin friction in turbulent boundary layers with severe pressure gradients. The gaps between the floating element and the surrounding wall are filled with a liquid in order to eliminate disturbing pressure forces on the element. The resulting friction forces are measured with piezo-electric transducers with high sensitivity and extremely small element displacement.

Skin friction measurements were taken in the turbulent boundary layer of a wind tunnel with circular cross-section at M [les ] 0·25. Severe adverse pressure gradients were generated by means of a step on the wall or, alternatively, by a conical centre body.

The new apparatus was mainly used to investigate the error of Preston tubes in adverse pressure gradients. It was necessary to develop a new measuring technique to improve the repeatability of the Preston tube readings.

The Preston tube error was found to depend on both the local pressure gradient P = (dp/dx) ν/ρ3τ and on the Preston tube diameter uτd/ν and to be independent of the upstream pressure distribution for the range of parameters covered by the experiments.

This paper presents exact solutions using toroidal co-ordinates to the equations of creeping fluid motion with the no-slip boundary conditions for a toroidal particle translating in a direction normal to the axis of symmetry or rotating about an axis normal to the axis of symmetry through an otherwise infinite expanse of quiescent fluid. The associated resisting force and resisting torque are computed for toroids of various geometrical ratios b/a, b being the smallest radius of the open hole and (b + 2a) being the radius to the outermost rim of the torus. These results are compared with approximate calculations based on slender-body theory and on the theory for interacting beads. The exact and approximate calculations become asymptotically equal as b/a becomes very large, but departures from the exact calculations are apparent for b/a less than 10−100 depending on the mode of motion and the method of approximation and the approximations are unreliable for b/a less than 2·0.

The linear Rayleigh–Taylor stability of superposed viscous fluids with interfacial transfer of mass and heat is first considered for layers of finite thickness. A dispersion relation is obtained. It is then employed to derive stability and instability criteria for the case of two semi-infinite layers as well as the case where one of the layers is finite. From these criteria one arrives at a critical dispersion relation and a new critical wavenumber. This new critical wavenumber is distinct from the classical value owing to the presence of a parameter which depends, in a very simple manner, upon the kinematic viscosity of the fluids, the surface tension and the rate of interfacial transfer of mass and energy. Also it is found that the stabilizing effect of the surface tension is neither affected by the arrangement of the system nor the direction of the temperature gradient. However, the effects of the viscosity and the gravity will depend upon the relative positions of the superposed fluids and the direction of the temperature gradient at the interface.

Hydrodynamic measurements were made with a triaxial hot wire in the full-coverage region and the recovery region following an array of injection holes inclined downstream, at 30° to the surface. The data were taken under isothermal conditions at ambient temperature and pressure for two blowing ratios: M = 0·9 and M = 0·4. (The ratio M = ρjetUjet/ρ∞U∞, where U is the mean velocity and ρ is the density. Subscripts jet and ∞ stand for injectant and free stream, respectively.) Profiles of the three mean-velocity components and the six Reynolds stresses were obtained at several spanwise positions at each of five locations down the test plate.

In the full-coverage region, high levels of turbulence kinetic energy (TKE) were found for low blowing and low TKE levels for high blowing. This observation is especially significant when coupled with the fact that the heat transfer coefficient is high for high blowing, and low for low blowing. This apparent paradox can be resolved by the hypothesis that entrainment of the mainstream fluid must be more important than turbulent mixing in determining the heat transfer behaviour at high blowing ratios (close to unity).

In the recovery region, the flow can be described in terms of a two-layer model: an outer boundary layer and a two-dimensional inner boundary layer. The inner layer governs the heat transfer.

Hydrodynamic data are reported in the companion paper (Yavuzkurt, Moffat & Kays 1980) for a full-coverage film-cooling situation, both for the blown and the recovery regions. Values of the mean velocity, the turbulent shear stress, and the turbulence kinetic energy were measured at various locations, both within the blown region and in the recovery region. The present paper is concerned with an analysis of the recovery region only. Examination of the data suggested that the recovery-region hydrodynamics could be modelled by considering that a new boundary layer began to grow immediately after the cessation of blowing. Distributions of the Prandtl mixing length were calculated from the data using the measured values of mean velocity and turbulent shear stresses. The mixing-length distributions were consistent with the notion of a dual boundary-layer structure in the recovery region. The measured distributions of mixing length were described by using a piecewise continuous but heuristic fit, consistent with the concept of two quasi-independent layers suggested by the general appearance of the data. This distribution of mixing length, together with a set of otherwise normal constants for a two-dimensional boundary layer, successfully predicted all of the observed features of the flow. The program used in these predictions contains a one-equation model of turbulence, using turbulence kinetic energy with an algebraic mixing length. The program is a two-dimensional, finite-difference program capable of predicting the mean velocity and turbulence kinetic energy profiles based upon initial values, boundary conditions, and a closure condition.

Laboratory experiments were conducted to measure the surface elevation probability density function and associated statistical properties for a wind-generated wave field. The laboratory data together with some limited field data were compared. It is found that the skewness of the surface elevation distribution is proportional to the significant slope of the wave field, §, and all the laboratory and field data are best fitted by \[ K_3 = 8\pi\S, \] with § defined as ($(\overline{\zeta^2})^{\frac{1}{2}}/\lambda_0 $, where ζ is the surface elevation, and λ0 is the wavelength of the energy-containing waves. The value of K3 under strong wind could reach unity. Even under these highly non-Gaussian conditions, the distribution can be approximated by a four-term Gram-Charlier expansion. The approximation does not converge uniformly, however. More terms will make the approximation worse.

The decay of a jet discharging from a circular nozzle parallel to and displaced from a solid surface is investigated under conditions where the transitional process from circular-jet flow to oblate wall-jet flow begins in the initial, transition or self-preserving regions of the original jet. The influence of displacement of the nozzle from the plane on the developed three-dimensional wall jet downstream is demonstrated and it is found that the transitional interaction with the plane is more extended when the plane interacts first in the initial zone of the circular jet. Measurements of turbulence and Reynolds stress show the transverse mixing parallel to the plane to exceed that perpendicular to the plane, and are generally consistent with the spreading rates in these two directions, the ratio of which approaches 8·5 at large distances from the nozzle. It is shown that the interaction between the plane and jet involves a relatively large-scale coherent motion in which components of velocity directed towards or away from the surface are associated with outflow or inflow along the surface. This motion is more extended in the direction parallel to the surface and provides a mechanism for the increases in mixing rate in the direction parallel to the plane.