It is vital to develop new engine technologies to dramatically reduce harmful emissions. The next generation of gas-turbines have increased efficiency through the increase of the overall pressure-ratio of the compressor, which will be a significant stride towards the reduction of emissions.
These high pressure-ratios present several challenges for designers. The higher the pressure ratio, the smaller the compressor engine core and blade height. This will affect the blade tip clearance, which is a vital factor in maintaining efficiency and reliability. The calculation of clearance values requires the prediction of thermal growth of the compressor discs, which requires knowledge of their temperatures and heat transfer coefficients. However, flow inside the air-filled cavity between compressor discs under these conditions is buoyancy induced. This creates a conjugate problem, where the heat transfer coefficients depend on the temperature distribution of the discs, and the temperature depends on the heat transfer coefficients. Furthermore, the fluid in the rotating cavity is three-dimensional, unsteady, and unstable. This is a result of Coriolis forces creating cyclonic and anti-cyclonic circulations, which produces circumferentially anti-axisymmetric flow. These flow features are known as 'structures' and are a result of the shroud and disc temperatures being greater than that of the air in the cavity. This temperature difference drives buoyancy forces, as well as causes a temperature rise in the axially flowing air through the bore of the compressor.
The research into this problem involves three major elements: experimentation, simulation, and theory. This PhD revolves around theory and will focus on mathematical modelling of the conjugate heat problem for individual rotating compressor discs. This will be carried out in parallel with experimental and computational research.
An existing model exists for closed cavity compressors discs (where the axial throughflow is separated from the air-filled cavity) in steady state. The first step of this research is to extend this model to transient conditions. A model for an open cavity compressor disc will then be developed, for both steady-state and transient conditions.
The models will require experimentally derived heat flux values on the compressor discs, which are determined from temperature measurements. The calculation of these fluxes is an inverse problem, where small uncertainties in the temperatures can lead to unmanageable uncertainties in the calculated heat fluxes. The method of Bayesian statistics, solved using neural networks, will be applied to reduce these uncertainties. These results may then be used to validate and inform the theoretical models.