Unsteady RANS modelling of wake–blade interaction: computational requirements and limitations

• Published in: Computers & fluids 2005, Vol.34 (1), p.3-21
• Main Authors: Lardeau, S., Leschziner, M.A.
• Format: Article
• Language: English
• Subjects: Applied fluid mechanics; Computational methods in fluid dynamics; Exact sciences and technology; Fluid dynamics; Fundamental areas of phenomenology (including applications); Hydraulic and pneumatic machinery; Physics; Rotational flow and vorticity; Separated flows
• ISSN: 0045-7930; 1879-0747
• DOI: 10.1016/j.compfluid.2004.04.001

An investigation is presented of the ability of the unsteady RANS approach to represent, from a numerical point of view, the characteristics and evolution of moving wakes as they interact with the flow around a turbine blade and in a related variable-passage configuration. The context of the study is wake-induced transition in unsteady turbomachine flows. Its emphasis is on numerical and computational issues, with physical aspects of predicting unsteady transition in the boundary layer of a turbine blade being covered in a companion paper. Issues addressed include: resolution and time-step requirements––especially in relation to the convective transport of scalar-wake properties, the sensitivity of the solution to the numerical scheme, and the relative merits and limitations of either including the wake-generating moving devices––in the present case an array of circular bars––as part of the computational process, or of prescribing the moving wakes as part of the flow-inlet conditions upstream of the blade passage. In relation to the latter option, the viability of extracting the wake conditions from a precursor RANS computation for an isolated wake-generating body is examined. In total, results for 21 test conditions are reported in an effort to derive some general conclusions and recommendations. The investigation demonstrates that (i) extremely high resolution is needed to ensure that numerical issues do not obscure physical phenomena and turbulence-model capabilities; (ii) the inclusion of the wake-generating device in the computation poses serious problems associated with vortex shedding; and (iii) the results of precursor calculations for the wake can be seriously flawed because of inherent limitations in the RANS approach.