The effects of wear on abradable honeycomb labyrinth seals

This thesis reports on work undertaken to understand the effects, due to wear, on the performance of abradable honeycomb labyrinth seals. The phenomena studied are aerodynamic in nature and include compressible flow, turbulent flow, recirculation and separation at a range of pressure ratios from 1.20 up to 3.50. Four primary methods of investigation were used: experimental, numerical using CFD, numerical using theoretical derivations and numerical using established labyrinth seal specific computer codes. Effects of seal clearance, pressure ratio and tooth to groove location have been investigated with overall performance and inter-seal pressure distribution recorded experimentally and numerically for comparison. Worn experimental results, when compared to their unworn equivalent, recorded large increases in mass flow of up to 50% when the labyrinth teeth are located centrally in the groove. Significant performance enhancements were achieved through offsetting the teeth with respect to the groove, particularly in an upstream sense. There was a marked deterioration with the labyrinth teeth located at the groove exit. Inter-seal pressure distributions showed that the first and final teeth did most of the work achieving significantly larger pressure drops which goes against current seal understanding of increasing pressure drop through the seal. Numerical work was undertaken to further investigate these effects. However, due to the complex 3-D geometry of an abradable honeycomb labyrinth seal a 2-D simplification technique was developed to speed up the investigative process. Using this technique CFD was found capable of replicating the experimental data regarding overall seal performance and inter-seal pressure distributions. The pressure on the final tooth proved to be the hardest experimental data to recreate using CFD, particularly at high pressure ratios when shocks are likely to form. Further numerical work was undertaken using computer codes and theoretical derivations. This work proved that the understanding of the seal loss coefficients used by both methods was not adequate for the current study with the experimental data recreated least successfully. Suggestions are given for enhancement of seal design, including axial location and seal computational routines, which will limit the impact of a 1.5% increase in operational cost that is likely to accrue from seal deterioration.