Computational investigation of torque on spiral bevel gears

This thesis describes the development of a numerical modelling strategy for simulating the flow around a shrouded spiral bevel gear. The strategy is then applied to a series of parametric variations of key shroud parameters. The shroud and gear in question are generic, although based upon those employed in the internal gear box of a Rolls-Royce aeroengine. The need to shroud the gear comes from the fact that a spiral bevel gear, when rotated, acts like a fan. Work is done by the gear to move the surrounding fluid, usually air with oil particles suspended in it, which creates a parasitic loss, referred to as the windage power loss. The work within this thesis is part of a larger project which has investigated how windage power loss can be affected by geometric features of gears and shrouds. This is important as for large diameter (>200mm) bevel gears running at high speeds (>10,000 RPM) the windage power loss forms a substantial part of the total power loss. The modelling strategy has been developed in this work by studying 4 different fluid flow settings: Taylor-Couette flow, Conical Taylor-Couette flow, an unshrouded spiral bevel gear, and a shrouded spiral bevel gear. Work on Taylor-Couette flow provided a basic setting in which to trial various numerical techniques and gain familiarity with the commercial CFD program which would be used throughout this thesis (FLUENT), along with the meshing program GAMBIT. It gave an understanding of the flow, which was then used to simulate the flow in a modification of Taylor-Couette flow where the cylinders are replaced with cones, called Conical Taylor-Couette flow. Comparisons were made between 4 popular turbulence models, allowing a decision to be made on the `best' turbulence model to use in the modelling of a shrouded gear, and to start to develop the strategy. This strategy was then applied to the more complex geometry of an unshrouded gear, simulating experimental data which had been created on an in-house rig. To confirm the applicability of the strategy to modelling shrouded spiral bevel gears, it was applied to two shrouds for which experimental data was available. It showed that numerical modelling can capture the relative performance of the shrouds well. The work then continued by considering a series of parametric variations, whereby 3 key shroud parameters are each varied in 3 manners, producing 27 variations. Each of these parameters can affect the windage power loss: an assessment of how much each parameter affects windage power loss has been given. A description of the flow field in `good' and `bad' cases has been given, and through approximating the flow by using the compressible form of Bernoulli's equation, reasons for a `bad' shroud being `bad' have been presented