Abstract
Various turbulent models (Spalart-Allmaras, Standard k-ε, RNG k-ε, Realizable k-ε and Reynolds Stress Models) along with Standard, and two-zonal wall functions are used to simulate inductively coupled plasma flows. The computational results can be classified into two categories: All the turbulent models that include low Reynolds number effects, such as, Low Reynolds number k-ε model, Spalart-Allmaras one-equation model, Standard k-ε model with two-zonal wall function, RNG model with turbulent viscosity determined by a differential equation, RSM etc., give similar modelling results. These models predict almost the same temperature contours which are similar to the one predicted by laminar model. The viscosity ratios in plasma region predicted by these models are very close to zero except for in the wall-neighbouring cells, which means the plasma flow is almost laminar. The other category contains those models that do not include the low Reynolds number effects, such as Standard, RNG and so-called Realizable k-ε models with standard wall function. They predict the plasma flow to be turbulence-dominated. In comparison with the results of experimentally measured heat fluxes to a substrate, the heat fluxes predicted by these models that include low Reynolds number effect are very close to experimental measurements while these models that do not include low Reynolds number effects deviate greatly from experimental measurements. It is found that the Reynolds stress model(RSM) appears to be the best predictive model.