Compressor aerodynamics at negative incidence

In the last decades, customers and regulators are continuously imposing more stringent requirements on safety, emissions and fuel consumption, pushing the manufacturers to optimise every single phase of the engine operations. Many goals in design point operations and off-design ones have already been reached. However, in order to meet the requirements on ground start and altitude relight capabilities, the sub-idle regime performance still needs to be fully comprehended. In this regime, the engine is operating at a speed below the idle one and, since these are extreme off-design conditions, innovative approaches to create high-fidelity prediction tools are required. For this purpose, one of the main challenges is to further understand the compressor behaviour. Many performance tools are available to predict it near design point, but, due to the lack of experimental data and the complexity of the flow, they are not reliable for sub-idle conditions. Hence, in the previous years, in Cranfield University a surrogate model for loss and deviation in sub-idle conditions was developed to be implemented in mean-line or through-flow codes. The maps generated through these tools will then be used for the whole engine sub-idle performance analysis. The main target of this research was then to improve and generalise this model by considering the effects of the camber angle and of the flow unsteadiness. Using the same 2D CFD parametric model, different blade geometries were considered. The results are suggesting that the camber angle has no effect on the flow deviation. For the pressure loss coefficient, a clear trend was not found probably due to the flow unsteadiness. Hence, transient simulations were performed, assessing the behaviour of the employed turbulence models as well. For the deflection case, the results were confirming the steady-state outcomes in all the cases. Regarding the pressure loss coefficient, there was a discrepancy between the data: the SAS-SST model behaved in a different way depending on the considered geometry, probably due to its scale-adaptive capacity, that in case of a parametric model makes impossible to have a fair comparison. However, the model has been used in the mean-line code, generating compressor maps that were successfully validated against experimental data.