Dimensionality reduction of non-buoyant microconfined high-pressure transcritical fluid turbulence

Abstract

This work utilizes a novel data-driven methodology to reduce the dimensionality of non-buoyant microconfined high-pressure transcritical fluid turbulence. Classical dimensional analysis techniques are limited by the non-uniqueness of scale-free groups and the lack of a general strategy for quantifying their importance. Instead, the data-driven approach utilized is based on augmenting Buckingham’s π theorem with ideas from active subspaces to overcome these limitations. Through this methodology, a principal dimensionless group has been identified that efficiently describes the behavior of the system in terms of normalized bulk turbulent kinetic energy. Additionally, a simplified version of the new dimensionless group is proposed, which presents the structure of a Reynolds number augmented with dynamic viscosity, thermal conductivity, or equivalently Prandtl number and isobaric heat capacity, and specific gas constant to account for thermophysical effects. Finally, the results obtained in this study, which is based on a realistic regime inspired by nitrogen at high-pressure microfluidic conditions, can be generalized to other fluids using the principle of corresponding states.