Vortical structures and pressure pulsations in draft tube of a Francis-99 turbine at part load: RANS and hybrid RANS/LES analysis

A. A. Gavrilov, A. V. Sentyabov, A. A. Dekterev, K. Hanjalic

Research output: Contribution to journalArticleScientificpeer-review

23 Citations (Scopus)

Abstract

Recognizing the limitations of the conventional linear-eddy-viscosity (LEVM) Reynolds-averaged Navier–Stokes (RANS) models to reproduce complex three-dimensional unsteady flows in hydraulic machinery, we performed a comparative assessment of a second-moment (Re-stress model, RSM) RANS closure and a hybrid RANS/LES method in capturing the flow and vortical structures in the draft tube of a Francis hydroturbine at off-design conditions. Considered is a case of part load (PL) at a flow rate of only 35% of the best efficiency point (BEP) characterised by multiple unsteady vortex systems. Despite some remaining uncertainties in generating the inflow conditions, both approaches reproduced reasonably well the measured mean velocity and the rms of its fluctuations, as well as the pressure spectrum with peaks detecting the precessing vortex core. In contrast to the common LEVMs, the Re-stress closure showed sufficient receptivity to intrinsic unsteadiness and reproduced well the overall flow and vortical patterns as well as the associated pressure pulsations in accord with the experiments. The hybrid RANS/LES method gave similar predictions as the RSM, but resolving a wider range of scales, which however, showed no significant effect on the dynamics of the dominant processing vortex core and the pressure pulsations.

Original languageEnglish
Pages (from-to)158-171
Number of pages14
JournalInternational Journal of Heat and Fluid Flow
Volume63
DOIs
Publication statusPublished - 1 Feb 2017

Keywords

  • Hybrid RANS-LES
  • Hydroturbine
  • Second-moment closure
  • Vortex rope

Fingerprint Dive into the research topics of 'Vortical structures and pressure pulsations in draft tube of a Francis-99 turbine at part load: RANS and hybrid RANS/LES analysis'. Together they form a unique fingerprint.

Cite this