Numerical and neural study of flow and heat transfer across an array of integrated circuit components

The present study is concerned with the flow and heat transfer for laminar flow over an array of rectangular blocks. These blocks represent finite heat sources on parallel plates (e.g. the components of integrated circuits). The study is based on simulations of three and six aligned hot small blocks. The investigation aims to determine the effect of different parameters (e.g. Reynolds number and blocks arrangement) on the magnitude and location of the maximum temperature at the surfaces of the hot blocks. The numerical investigation was carried out using the commercial code ANSYS 5.4 based on the finite element technique. Also, experimental investigation was carried out for verification and more understanding of the problem. Neural networks were utilized to predict the values of the mean Nusselt number of the integrated circuit components. Useful discussions and fruitful comments are presented.