This CFD study compared multiple turbulence models against experimental data for a 0.29 m Stairmand cyclone. Appropriate discretisation improved some EVMs, and realizable k-epsilon with curvature correction matched linear/quadratic pressure-strain RSM mean-flow accuracy under the studied conditions with less computation. It challenges routine RSM preference but does not cover every cyclone or particle regime.
Key findings
- Some unmodified EVMs improved with appropriate discretisation where curvature effects were weak. Curvature correction improved EVM mean-flow prediction, and corrected realizable k-epsilon approached two RSMs with shorter computation. The authors caution against uncritical interpretation of RANS-predicted RMS velocity fluctuations.
Why this matters globally
Cyclones serve pollution control, energy and processing worldwide. Faster adequate models can reduce optimisation cost, while poor choices can mispredict collection and pressure drop.
Thai researcher contribution
KMITL and Chulalongkorn University researchers used experimental reference data to reassess CFD practice rather than automatically preferring the most complex model.
Limitations to consider
Conclusions derive from one cyclone geometry and operating regime, focusing on mean flow rather than polydisperse collection, erosion or transient coherent structures. Runtime depends on mesh, solver and hardware. Agreement with RSM does not guarantee every output is correct.