Optimized design of axial turbine rotor blades considering a wide range of attack angles

Abstract

To extend the available range of attack angles and reduce profile loss, this paper develops a systematic aerodynamic optimization methodology for turbine blade profiles. This methodology integrates Latin Hypercube Sampling (LHS), Artificial Neural Networks (ANN), and Genetic Algorithms (GA). A multi-condition optimization was performed on the PACK-B profile by considering four key geometric parameters: leading edge radius, leading edge wedge angle, trailing edge radius, and trailing edge wedge angle, at attack angles of 0°, 10°, 20°, and −30°. The results demonstrate a significant enhancement in the profile’s aerodynamic performance. The findings indicate that the optimization at the 10° attack angle yielded the most pronounced improvements. The optimized profile achieved a substantial extension of its usable attack angle range to −40° to 20°, accompanied by a remarkable reduction in profile loss of up to 49%. Parameter sensitivity analysis revealed that the leading edge radius has the most significant influence on profile loss, followed in descending order by the trailing edge radius, trailing edge wedge angle, and leading edge wedge angle. A comprehensive comparison shows that an optimization strategy targeting positive attack angle conditions effectively improves performance at both the design and off-design conditions. This approach successfully broadens the operational attack angle adaptability and minimizes losses, thereby providing a valuable theoretical foundation and an engineering-applicable methodology for the design of high-performance turbine blades.