Thermal and flow characteristics of buoyancy-driven non-Newtonian flows at a high Rayleigh number of 10⁷ and predictions from an artificial neural network

The thermal and flow characteristics at Ra = 107 were evaluated in a square cavity containing a circular cylinder in different places along the diagonal and horizontal centerlines. The enclosure contained non-Newtonian fluids of pseudoplastic and dilatant natures. The power-law index was varied in the range of 0.6–1.6 with an interval of 0.2 and a fixed Prandtl number of 10. The effects on the laminar natural convection are reported. The flow regimes were categorized as steady symmetric, steady asymmetric, non-periodic unsteady symmetric, non-periodic unsteady asymmetric, periodic unsteady symmetric, and periodic unsteady asymmetric. Artificial neural network was used to predict the thermal performance in the enclosure. The thermal transport in cases of n < 1 was much higher than that in cases of n > 1. The ANN model was effective in estimating the heat transfer performance with appropriate training.