Robust design of film-cooled high-pressure turbine blades in the face of real geometric variations

Abstract

The aerothermal performance of turbine blades is negatively affected by unavoidable variations, for example, those due to the manufacturing process. No blade ever exactly conforms to its nominal design intent geometry and variations, even slight ones, can have detrimental effects on aerodynamic performance, blade temperatures and blade lifespan. As opposed to conventional deterministic design, robust design is one option to tackle the problem posed by deviations by incorporating them directly into the design process. In robust design, both performance mean and scatter can be optimized concurrently to make blades less sensitive. Such a workflow, for the first time based on real variations extracted from 58 digitized 3D turbine blade scans from two turbofan engines, is presented and applied to aerodynamically optimize an industrial turbine rotor. The approach uses high-fidelity geometric models, a non-intrusive uncertainty quantification method and efficient robust optimization to effectively locate Pareto-optimal robust designs. One selected design is validated and shown to be desensitized to the underlying observed manufacturing variability, which is proven to be crucial to obtain realistic results. In addition, the first probabilistic assessment of film cooling hole variations on blade performance is also presented. An accurate reverse engineering workflow for blade shape and film cooling hole variations is used. Particular focus is placed on the trailing edge and the trailing edge film cooling holes due to their impact on performance, for example on capacity. It is shown that positional deviations of film cooling holes from nominal are largely due to blade shape variations as opposed to other sources. Therefore, a new model for film cooling variability termed “virtual film cooling hole manufacture”, which replicates the real manufacturing process of the holes, is introduced. The methodology essentially exploits the finding that blade shape variations are a major cause of film cooling hole variations. Through a comparison with an uncooled setup at the same operating point, it is also shown that trailing edge film cooling hole variability itself is important, especially for capacity settings. The analysis model choice, either continuous film cooling strips with uniform flow, discrete holes or a combination of both, impacts uncertainty quantification options. The presented workflow enables comprehensive uncertainty quantification and robust design optimization of film-cooled turbine blades with fully-featured geometric models and includes realistic variations of the entire blade shape, of the shroud and of all film cooling holes. Artificially constructed blade models are shown to exhibit the same performance variation as the original measured sample of blade scans and can be used to compute output blade performance statistics for uncertainty quantification.

More information

Identifiers

ORCID for Andrew Keane: orcid.org/0000-0001-7993-1569

Catalogue record

Export record