Akademik Veri Yönetim Sistemi
International Journal of Heat and Fluid Flow, cilt.112, 2025 (SCI-Expanded)
This study presents the application of an artificial neural network (ANN) model to predict the Nusselt number of CuO-water nanofluid in a rectangular grooved channel under the effect of a magnetic field. In the developed ANN model, while Reynolds number (250 ≤ Re ≤ 1250), volume fraction of nanofluids (0 ≤ Φ ≤ 5), and Hartmann numbers (0 ≤ Ha ≤ 28) were taken as input parameters, Nusselt number was selected as the output parameter. Data were generated from a computational fluid dynamics (CFD) code by discretizing equations using the finite difference method. Therefore, the outcomes acquired from numerical simulations using CFD code were used for training and testing the generated ANN model. According to the results the generated ANN model can accurately predict the Nusselt number with a mean absolute percentage error (MAPE) of 0.4288 %, mean absolute error (MAE) of 0.0351, and root mean square error (RMSE) of 0.0540 in testing and of 0.3177 % MAPE, 0.0249 MAE and 0.0328 RMSE in the training. Furthermore, the correlation coefficient (R) values are observed as 0.9998 and 0.9988 in training and testing phases, which demonstrate the prediction success of the generated ANN model. Notably, the ANN model reduced computational time from 8 h, using CFD methods, to just 10 min for testing cases, showcasing its efficiency in handling nonlinear flow cases where traditional CFD methods may struggle. This study represents a novel contribution to the field as one of the first to apply ANN techniques for predicting heat transfer in grooved channels under magnetic fields and nanofluid flow, offering potential applications in the design of thermal systems in industries such as electronics cooling, nuclear reactors, and metallurgy.