Application of Artificial Neural Network for the Prediction of Pool Boiling Heat Transfer Coefficient of Nanofluid with External Magnetic Field Effect
Abstract
Analyzing the pool boiling heat transfer of nanofluids with artificial neural network (ANN) is the primary goal of this research, which also intends to conduct experimental investigations. The trials were conducted with magnetic fields ranging from 0.0008 to 0.0012 Tesla and nanofluid concentrations from 0.015 to 0.075 vol%. The pressure inside the chamber might be anything from 50 to 150 kPa. Before conducting an ANN analysis, a training session was conducted to teach the network how to estimate the pooled boiling heat transfer performance of mixes containing nanoparticles of TiO2 and refrigerant R141b in the presence of a magnetic field. There is evidence that the strength is greatly impacted by particle Brownian motion, which results in improved heat transfer. The heat transfer coefficient is greatly affected by pressure. The structure of the ANN model consists of four input neurons, two hidden layers, and one output neuron. Seventy percent of the 468 datasets are used for training, while the remaining 15% and 15% are divided equally between testing and validation. Given that the root mean deviation was only ±5%, it was enough for sensitivity. This method reduces the network's training time while increasing calculation efficiency. This architecture performs when training neural networks for classification and forecasting tasks that need accuracy and efficiency in tuning.