Metal foam is a new class of material, which presents application in several different fields of interest. For this paper, aluminium foam was tested in order to evaluate the heat exchange peculiarity, field where this material excel due to great surface for unit volume. Objects of the present paper is the construction of a prototypal heat exchanger composed of aluminium foam, with the aim to evaluate the performance of the foam in heat exchange. Foam was electrodeposited at a later stage with copper in order to increase the foam conductibility. The unpredictable behaviour of metal foam in heat exchange require a great control of device operation. Further purpose of this paper is the implementation of two neural networks with resilient backpropagation algorithm, NET1 e NET2, trained with results obtained in the experimental test executed. These neural networks are orientated towards prediction of metal foam heat exchanger behaviour and aimed to evaluate the performance in terms of heat exchange coefficient and efficiency. Comparison between simulated and predicted value, shows that the networks created allow giving a generalization of the problem and finding a correlation correlation between all the parameters considered.
Baiocco, G., Tagliaferri, V., & Ucciardello, N. (2017). Neural Networks Implementation for Analysis and Control of Heat Exchange Process in a Metal Foam Prototypal Device. In Procedia CIRP (Vol. 62, pp. 518–522). Elsevier B.V. https://doi.org/10.1016/j.procir.2016.06.035