Abstract:
Problems related to phase change and mass transfer are characterized by high nonlinearity, moving boundaries and sharp
changes in parameters, which complicates their numerical solution by traditional methods. The aim of this work is to
study the possibility of using a new method Physics Informed Neural Networks, which uses neural networks to approximate
unknowns, to solve such problems. The method was applied to solve Stefan problems for one and two phases, as well as to
numerically analyze the problem of the motion of a gas bubble surrounded by a liquid. The method demonstrated good
agreement with other solutions for Stefan problems and made it possible to simulate the bubble motion, although with
some errors. There is significant potential for further development of this method for solving heat and mass transfer
problems.
Key words:phase change, mass transfer, Stefan problem, neural networks.
Fulltext:
PDF file (1607 kB)