A Path Towards the Aerodynamic Robust Design of Low Pressure Turbines

Airline companies are continuously demanding lower-fuel-consuming engines and this leads to investigating innovative configurations and to further improving single module performance. In this framework the Low Pressure Turbine (LPT) is known to be a key component since it has a major effect on specific fuel consumption (SFC). Modern aerodynamic design of LPTs for civil aircraft engines has reached high levels of quality, but new engine data, after first engine tests, often cannot achieve the expected performance. Further work on the modules is usually required, with additional costs and time spent to reach the quality level needed to enter in service. The reported study is aimed at understanding some of the causes for this deficit and how to solve some of the highlighted problems. In a real engine, the LPT module works under conditions which differ from those described in the analyzed numerical model: the definition of the geometry cannot be so accurate, a priori unknown values for boundary conditions data are often assumed, complex physical phenomena are seldom taken into account, operating cycle may differ from the design intent due to a non-optimal coupling with other engine components. Moreover, variations are present among different engines of the same family, manufacturing defects increase the uncertainty and, finally, deterioration of the components occurs during service. Research projects and several studies carried out by the authors lead to the conclusion that being able to design a module whose performance is less sensitive to variations (Robust LPT) brings advantages not only when the engine performs under strong off-design conditions but also, due to the abovementioned unknowns, near the design point as well. Concept and Preliminary Design phases are herein considered, highlighting the results arising from sensibility studies and their impact on the final designed robust configuration. Module performance is afterward estimated using a statistical approach.