CSE Community Seminar

March 7, 2025, 12-1PM

Wrong but not guilty: the Reynolds analogy and model error in CFD

Emilio Baglietto
Associate Department Head and Professor,
Nuclear Science and Engineering, MIT

Abstract:

Professor Kasagi insightfully noted in 1989 that “Modeling of turbulent heat transport is apt to rely on intuition and speculation rather than direct experimental evidence”. This reliance has led the engineering community to widely adopt the simple Reynolds analogy, which equates turbulent momentum and heat transfer processes. However, this approach has been criticized by a significant portion of the turbulence modeling community, who argue that the Reynolds analogy introduces substantial model errors in heat transfer calculations, particularly in non-unity Prandtl number fluids. This critique has spurred years of research aimed at improving turbulent heat flux models, including my own efforts, though with limited success. I will challenge this narrative, and propose that the inadequacies of turbulence models in predicting turbulent viscosity and complex unsteady phenomena are the main culprits. This conclusion is based on first principles and empirical evidence from bulk mixing flows and is further supported by novel Direct Numerical Simulation (DNS) data obtained through extensive collaborative efforts. Our revised understanding of turbulent heat flux underscores the need for enhanced turbulence models, including hybrid models, and/or the implementation of model error quantification techniques. These strategies aim to direct the modeling community’s focus towards the most critical sources of model error and bolster confidence in CFD heat transfer simulations, which is essential for the broader adoption of CFD in mission-critical applications.

March 7, 2025, CSE Community Seminar
Emilio Baglietto
Associate Department Head and Professor
Nuclear Science and Engineering, MIT