Title: 

Scientific Machine Learning for Transport Phenomena in Thermal and Fluid Sciences

Abstract:

Engineering design optimizations using fluid dynamics simulations can be computationally expensive to the extent that they can limit the scope of design space exploration. Physics-based machine learning techniques have been proposed as an alternative approach for data-driven inverse modeling and optimization problems involving partial differential equations, but current techniques have severe limitations. To this end, the proposed research aims to develop a new machine learning framework that can satisfy the underlying governing equations and their boundary conditions much more accurately than the existing machine learning-based approaches. Furthermore, we propose to extend this new approach to facilitate the learning of large-scale incompressible fluid flow problems using domain decomposition methods.