Abstract

High-fidelity computations of turbulent flows at high-pressure supercritical fluid conditions present significant challenges. Besides the inherent broadband nature of the flow, the rapid variation of thermophysical properties across the pseudo-boiling region can result in additional complexities in terms of strong localized density gradients, spurious pressure oscillations, non-linear behaviour of fluids, and amplification of aliasing errors. Different research groups have utilized distinct approaches to achieve numerical stability, mostly resorting to upwindbiased schemes, artificial dissipation and/or high-order filtering. However, in these strategies, stability is achieved at the expense of artificially suppressing part of the turbulent energy spectrum. In this regard, this work aims to explore the suitability, in terms of stability and accuracy, of recently proposed energy-preserving schemes for scale-resolving simulations of supercritical turbulence. For ideal gases, such type of methods have been demonstrated to provide stable and accurate computations of turbulence by preserving kinetic energy and/or other quantities of physical relevance. However, their extension to real-gas thermodynamic frameworks is still in its infancy, and consequently requires to be carefully investigated. To this objective, this work analyzes the performance of different classical and energy-preserving discretizations under ideal-gas conditions, and carries out an initial assessment of their performance at high-pressure supercritical fluid regimes. The results obtained indicate that their extension to real-gas thermodynamics is not straightforward, and consequently motivate the necessity to develop new solutions able to satisfy the desired stability and accuracy requirements.