The long time effect of weak rotation on an internal solitary wave is the decay into inertia-gravity waves and the eventual formation of a localised wavepacket. Here this initial value problem is considered within the context of the Ostrovsky, or the rotationmodified Korteweg-de Vries (KdV), equation and a numerical method for obtaining accurate wavepacket solutions is presented. The flow evolutions are described in the regimes of relatively-strong and relatively-weak rotational effects. When rotational effects are relatively strong a second-order soliton solution of the nonlinear Schrödinger equation accurately predicts the shape, and phase and group velocities of the numerically determined wavepackets. It is suggested that these solitons may form from a local Benjamin-Feir instability in the inertia-gravity wave-train radiated when a KdV solitary wave rapidly adjusts to the presence of strong rotation. When rotational effects are relatively weak the initial KdV solitary wave remains coherent longer, decaying only slowly due to weak radiation and modulational instability is no longer relevant. Wavepacket solutions in this regime appear to consist of a modulated KdV soliton wavetrain propagating on a slowly varying background of finite extent.

A procedure is described that develops the nondimensional design of a radial inflow turbine rotor. The design is developed, for any specified nondimensional power ratio, with the objective of minimizing the inlet and discharge Mach numbers so that the passage losses are minimized. Initially state-of-the-art efficiencies are assumed, but these are later modified through the specification of empirical losses. The resultant nondimensional design can be transformed to absolute dimensions through the specification of the inlet stagnation conditions and the mass flow rate of the working fluid.

An experimental study is presented which considers the flow structure in the vaneless volute of a radial inflow turbine. The data show evidence of significant secondary flows around the volute. The volute tested generated a uniform ‘free vortex’ type structure and produced a uniform mass flow distribution around the periphery of the rotor. The flow within the volute was turbulent although this did not appear to have a detrimental effect on the overall performance of the volute. However, the turbine performance appears to be insensitive to variations in the inlet conditions in the axial direction.

A unified one-dimensional analysis and design procedure for radial and mixed flow turbines is presented. This procedure considers the flow in all of the turbine components, including the inlet volute, nozzled or nozzleless inlet duct, rotor and exit duct. Losses are included for each component; in the stationary ducts, friction is the main loss mechanism, and in the rotor skin friction, clearance and disk friction are considered separately. Incidence losses at entry to both nozzle and rotor are calculated using one of two loss models. For each component, the procedure is to combine the equations of continuity, momentum, energy, and entropy into a single dimensionless expression for the Mach number at exit. This approach has proved possible under a wide variety of circumstances, and with its flexibility and ease of programming for solution by computer would seem to have many applications in one-dimensional ducted flow. The predicted performance of a radial inflow nozzleless turbine is compared with experimental results. The comparison is considered to be sufficiently good to confirm the validity of this approach. The results obtained with each of the two incidence loss models are compared. Each model is found to have its own advantages and fields of application.

A procedure to predict the complete performance map of turbocharger centrifugal compressors is presented. This is based on a one-dimensional flow analysis using existing published loss correlations that were available and thermodynamic models to describe the incidence loss and slip factor variation at flow rates which differ from the design point. To predict the losses within the complete compressor stage using a one-dimensional flow procedure, it is necessary to introduce a number of empirical parameters. The uncertainty associated with these empirical parameters is assessed by studying the effect of varying them upon the individual losses and upon the overall predicted performance.