Turbulence Modelling: Needs, Obstacles, and Opportunities

The persistent need to assess and improve RANS models, in this era of Turbulence-Resolving Simulations, is justified. The tendency to stagnation is partly explained by factors such as: the sheer scope of the project, given minimal tools and all the constraints of CFD solvers; the disappointing contribution of DNS; a newly argued Structural Limitation for "classical" RANS models; the lack of maturity in AI work; and the surprising prevalence of what are to the author obvious Fallacies. The user-adjustable models GEKO and SG26 are discussed. On a more positive note, the integration with experiment and DNS is now stronger thanks to modern flow-field measurements and creative quantities extracted from DNS.