The effect of purge-mainstream density ratio on turbine cavity flows.

A major challenge facing gas turbine designers is preventing hot (1300-1800K) main-annulus flows entering the wheel-space between the turbine disc (rotor) and its adjacent casing (stator). Designers tend to use complex rim seal designs and divert relatively dense sealant flows (800K) to avoid this occurring in the gas turbine. Insufficient sealing can cause hot gas ingress deep into the wheel-space and can lead to premature failure and limited life of the components, excessive sealing can cause reductions in efficiency of the engine which would lead to it becoming uncompetitive in a world of increasing fuel prices and emphasis on CO2 emissions. Preliminary studies of hot gas ingress conducted at Bath employed CO2 as a tracer gas at isothermal conditions (300K). Early results revealed a significant reduction in rim-seal effectiveness when the density ratio between the hot main flow and cold sealing flow was not properly modelled, demonstrating the potential for significant impact within industry.
This PhD research will primarily be conducted in the Large Annulus Rig (LAR) at the University of Bath and seek to investigate the effect of the density ratio between the purge flow and main annulus flow on ingress into the rotor-stator cavity of a geometrically scaled industrial turbine stage. This will be achieved through a number of individual considerations. Firstly, the LAR requires a new collection system, allowing for a fundamental study of density-ratio with insensitivity to exit swirl. The instrumentation available in the LAR will also be reconfigured to include high frequency pressure transducers. At this point, investigations into the density ratio will begin by varying the CO2 composition of the purge flow. To ensure this is consistent with simulations, bespoke CFD will be conducted in OpenFOAM to directly match the experiments. This concomitant approach will aim to unravel the complex ingress mechanisms and significant effect of density ratio. The CFD will then be further supplemented (and validated) by using Volumetric Velocimetry (VV) measurements to identify flow structures. The first aspect that will be investigated is the presence of vanes and blades on the density ratio effect. This will be achieved by conducting experiments and simulations with a simple, un-bladed ring on both the rotor and stator, then introducing turning vanes onto the stator, and blades onto the rotor. This will be modelled concurrently in CFD by having no vanes or blades, then either vanes or blades and finally both vanes and blades. The second stage will look into the effects of end-wall contouring on the density ratio. Finally, using the knowledge obtained, scaling models and equations will be developed which could be used by industry during the preliminary design phase when it would be uneconomical and impractical to use CFD due to the rapid development of designs.