Project title: Artificial Neural Networks in Computational Fluid Dynamics 
Supervisors: Dr Salvador Navarro-Martinez, Dr Stelios Rigopoulos
Artificial Neural Networks (ANNs) are becoming increasingly popular in the search for a solution to nonlinear systems in which the relationship between the input variables and the objective function is unclear. In this project, different ANNs architectures are explored to approximate closure models for isotropic and near-wall turbulence. The results show high correlation between predicted turbulent components and numerically resolved terms obtained from large eddy simulation, Gaussian filtered from direct numerical simulation. Successful evaluation of the trained network in different flow conditions validates the potential behind using ANNs to close turbulence. Should this project be developed further, a non-dimensional classifier could be developed to categorise the specific flow prior to the neural network. This could potentially benefit from additional flexibility, allowing the predictor to activate different networks based on parameters such as the Reynolds number and the normalised wall distance.