Heat transfer coefficient prediction of a porous material by implementing a machine learning model on a CFD data set
Avalos Gauna, Edgar
Palafox Novack, Leon Felipe
MetadataShow full item record
During many years, the search for new and improved materials has been an arduous task. It has mainly focused on experimentation and in recent years on computer aided techniques (i.e. numerical simulation). These two approaches defined the way material science works. Yet, both techniques have shown cost-efficiency disadvantages. Optimization algorithms, like the ones used in machine learning, have proven to be an alternative tool when dealing with lots of data and finding a particular solution. Even though the use of machine learning is a well stablished technique in other fields, its application in material science is relatively new. Material Informatics provides a new approach to analyse materials such as porous metals by employing previous data sets. This paper studies a new technique to predict the heat transfer coefficient of an open-cell porous structure while running water passes through the material. A CFD data set was employed by a Machine Learning technique in order to establish a relationship between the input parameters (porosity, pore size, pore distribution and flow rate) and the heat transfer coefficient of the sample. The results obtained from the analyses were compared with previous findings, concluding that by utilising a Machine Leaning technique is possible to obtain a more accurate and much better fit model. © 2019, Avestia Publishing.
The following license files are associated with this item: