Investigation of turbulence model selection on the predicted flow behaviour in an industrial crystalliser — RANS and URANS approaches

Gary J. Brown, David F. Fletcher, Jeremy W. Leggoe, David S. Whyte

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

An understanding of the flow behaviour in industrial crystallisers is critical to addressing mechanical design considerations such as impeller selection and off-bottom solids suspension. In the current study, Computational Fluid Dynamics (CFD) models were used to predict the flow field in a laboratory-scale alumina precipitator, with the objective of understanding which turbulence model can predict the time-averaged flow field in the vessel using the most efficient computational approach. The study focusses on RANS and URANS approaches using the k-ε, RNG k-ε, SST, SSG Reynolds Stress and explicit algebraic Reynolds Stress models (EARSM), with the predictions compared with high quality Laser Doppler Velocimetry (LDV) data from a laboratory-scale vessel. The best agreement with experimental data is achieved using the k-ε and k-ε EARSM models, with all other models either under or over-predicting the reattachment height of the flow in the annulus between the draft tube and the outer vessel wall. Of the models tested, the k-ε EARSM model is also found to give the best predictions of the RMS velocity fluctuations in the vessel. The use of turbulence production limiters, such as Kato-Launder, in conjunction with the k-ε model is found to result in worse agreement with the experimental data. The SST model significantly over-predicts the reattachment height and it is found that the high eddy viscosity generated in the separating shear layers exiting the draft tube, together with the recirculated turbulence in this geometry, suppresses the correct behaviour of the SST blending functions. The Reattachment Modification to the SST model is found to reduce the reattachment height, but the results still show significant deviation from the experimental data.

Original languageEnglish
Pages (from-to)205-220
Number of pages16
JournalChemical Engineering Research and Design
Volume140
DOIs
Publication statusPublished - 1 Dec 2018

Fingerprint Dive into the research topics of 'Investigation of turbulence model selection on the predicted flow behaviour in an industrial crystalliser — RANS and URANS approaches'. Together they form a unique fingerprint.

Cite this