Physics informed neural networks for triple deck

Purpose This paper aims to introduce physics-informed neural networks (PINN) applied the two-dimensional steady-state laminar Navier–Stokes equations over a flat plate with roughness elements and specified local heating. The method bridges gap between asymptotics theory three-dimensional turbulent flow analyses, characterized by high costs in analysis setups prohibitive computing times. results indicate possibility of using surface heating or wavy control incoming field. Design/methodology/approach understanding mechanism is normally caused unsteady interactions aircraft structure flows as well some studies have shown, can significantly influence fluid dynamics inducing perturbations velocity profile. Findings description boundary-layer flow, based upon triple-deck structure, shows how generate an interaction inviscid region viscous near plate. Originality/value To best authors’ knowledge, presented approach especially original relation innovative concept PINN solver asymptotic viscous–inviscid boundary layer interaction.