HEAT TRANSFER CHARACTERISTICS OF SUPER CRITICAL CARBON DIOXIDE FLOW IN HORIZONTAL PIPES USING COMPUTATIONAL FLUID DYNAMICS AND ARTIFICIAL NEURAL NETWORKS

Abstract

In the 21st century, the demand for eco-friendly, affordable energy is on the rise, and nuclear energy is one of the most practical methods of meeting this demand. Due to the high energy produced in the form of heat in these reactors, the design of the heat exchangers used is very important, as a failure can be catastrophic. Recently, heat exchangers using super critical fluids are gaining interest due to their drastic property variations near the pseudo critical point. In the present work, the validation of a computational fluid dynamics model is first carried out and compared against the available turbulence models in the software. Out of the six turbulence models used, <i>k-&epsilon;</i> Std model with the enhanced wall treatment is found to predict the wall temperature and Nusselt number with the average error of 12&#37;. Later in the study, a detailed parametric analysis was carried out for a wide range of mass flux (20 kg/m²-s to 2900 kg/m²-s), heat flux (0.5 w/m² to 1000 w/m²) and operating pressure (7.5 MPa to 9.5 MPa). Total of 81414 data collected from this parametric study is processed by an Artificial Neural Network model, which is then trained to provide values for the top wall Nusselt number. An artificial Neural Network model created can predict the top wall Nusselt number for the entire range of operating parameters within the average deviation of 9&#37;.