Heat transfer characteristics for upward flow of supercritical water in a vertical pipe - Computational fluid dynamics analysis and artificial neural network prediction

Supercritical fluid flows can be studied effectively using Computational Fluid Dynamics (CFD) for analysis and Artificial Intelligence for prediction. The application of machine learning to fluid flow problems yields accurate results with minimal computational expenditure. In the current work, a numerical study on heat transfer characteristics of supercritical water in a vertically upward pipe flow has been carried out. A three-dimensional model with an inner diameter of 7.5 mm has been used to investigate the flow behaviour. The k - epsilon Re -Normalization Group (RNG) turbulence model is found to be well suited and has been used to carry out simu-lations. Parametric studies have been performed for a mass-flux range of 150-1650 kg/m2s, heat-flux range of 150-1650 kW/m2 and operating pressure range of 23-28 MPa. Buoyancy effects on heat transfer have been analysed for different operating conditions. Onset of heat transfer deterioration has been investigated for the considered parametric conditions. A correlation for local Nusselt number has been proposed using Artificial Neural Networks (ANN) with an absolute average relative deviation (AARD) of 1.46% using 83,619 data samples.