This paper presents local Nusselt numbers computed from experimental measurements of surface temperature of compressor discs in a multiple rotating cavity test rig with axial through-flow. A validated 2D steady state heat conduction analysis methodology is presented, using the actual test geometry, and 95% confidence intervals calculated using Monte Carlo simulation. Sensitivity of the solution to curve fitting types, geometric simplification and surface instrumentation are explored.
The results indicate that polynomial curves fits, whilst computational simple, are unsuitable especially at higher orders. It is shown that geometric simplifications, that typically simplify the algorithmic implementation, may also omit significant variation in heat flux at critical stress relieving locations. The effect of reducing measurement points in the analysis is to both over-predict heat transfer and increase the uncertainty of the results. Finally, the methodology is applied to previously published thermal data from the University of Sussex, facilitating qualitative discussion on the influence of the governing parameters.
Whilst this study does not overcome the inherent uncertainty associated with inverse solutions it is intended to present a methodology that is readily available to the wider community for the analysis of thermal test data and suggests some guidelines at the planning and post-processing stages.
The range of experiment reported here covers: 1.13 × 105 < Rez < 5.14 × 105, 1.65 × 106 < Reθ < 3.16 × 106, 0.10 < Ro < 0.60 and 3.40 × 1011 < Gr < 1.25 × 1012.