An artificial neural network approach to bifurcating phenomena in computational fluid dynamics

This work deals with the investigation of bifurcating fluid phenomena using a reduced order modelling setting aided by artificial neural networks. We discuss POD-NN approach dealing non-smooth solutions set nonlinear parametrized PDEs. Thus, we study Navier–Stokes equations describing: (i) Coanda effect in channel, and (ii) lid driven triangular cavity flow, physical/geometrical multi-parametrized setting, considering effects domain’s configuration on position bifurcation points. Finally, propose manifold-based diagram for non-intrusive recovery critical points evolution. Exploiting such detection tool, are able to efficiently obtain information about pattern flow behaviour, from symmetry breaking profiles attaching/spreading vortices, even advection-dominated regime.