Enhanced Heat Flux Closure from Data Driven Modeling

Accurate prediction of surface heat flux in turbulent flow fields is critical for the design of air breathing hypersonic flight vehicles, to inform efficient thermal protection systems, and mitigate in-flight deformations leading to scramjet engine unstart. Current workhorse modeling methods, such as Reynolds Averaged Navier-Stokes (RANS) with conventional turbulence closure models, cannot accurately characterize hypersonic flowfield features, especially near regions with strong shocks, shockwave-boundary layer interactions, and other flow phenomenology encountered during hypersonic flight. Recent developments in the field of artificial intelligence and machine learning (AI/ML) enable data-driven and physics-informed deep neural networks to model the turbulent heat transport closure, provide accurate predictions of surface quantities, and enhance computational design efforts for hypersonic flight vehicles. In this STTR effort, Spectral Sciences, Inc. and Virginia Tech will extend and apply their novel AI/ML training framework for enhanced, model-agnostic turbulent heat flux closure for Navy-relevant hypersonic air breathing vehicles.