On the solution of out of equilibrium flows with the P-DNS method

GIMENEZ, JUAN M.; LARRETEGUY, AXEL; NIGRO, NORBERTO; IDELSOHN, SERGIO

Paris

Congreso; 14th World Congress in Computational Mechanics; 2021

Accepting as a premise that the solution of a Direct Numerical Simulation (DNS) is a reliable result, Pseudo-DNS (P-DNS) is a multi-scale method where both the coarse and the fine scales are solved numerically. The most expensive part of the computations, the solution of the fine-scale, is performed offline where its results, stress tensors at equilibrium, are stored in a dimensionless database. In the coarse-scale solution, fine-scale stresses must be advected and a Lagrangian formulation throught massless particles is employed for that task. When solving flows out of equilibrium with P-DNS, not just the local current conditions but also the history of strain rates experienced by the particle must be taken into account, and in this work, an approach based on memory fluids is presented to tackle this issue. From a large set of fine-scale simulations where transitions between two equilibrium states are solved, it is found that the proper time scale for transient processes is inversely proportional to the local velocity gradient, which is in agreement with observations on the bibliography. The P-DNS method including this memory model is evaluated in some problems with flows out of equilibrium obtaining accurate results comparing with DNS solution, regardless of the level of discretization employed for the coarse-scale.