International Journal of Science, Engineering and TechnologyAuthors: Yogesh Chouksey, Nitin Shrivastava, Sunil Kumar
Abstract: Thermal performance prediction of compact heat sinks is important during early design stages of laptop cooling systems. While computational fluid dynamics (CFD) provides accurate evaluation of heat transfer behavior, repeated simulations are computationally intensive. In this study, a simple artificial neural network (ANN) model is developed as a surrogate tool to predict the surface-averaged Nusselt number of fins in a specific laptop heat sink design using CFD-generated data. A total of 30 data samples corresponding to six Reynolds numbers and five fin configurations were used for training and testing. Reynolds number and fin type were employed as input parameters, while the surface Nusselt number served as the output. The ANN was implemented using a single hidden-layer feedforward architecture. Model performance evaluated on the test dataset yielded a root mean square error (RMSE) of approximately 4.33 and a mean absolute percentage error (MAPE) of about 6.6%, indicating satisfactory predictive capability. The results demonstrate that the ANN successfully captures the relationship between Reynolds number, fin geometry, and convective heat transfer performance, providing a computationally efficient approach for preliminary thermal assessment of the specified heat sink configuration.
DOI: http://doi.org/10.5281/zenodo.18639815
