Simulation and analysis of turbulent flows

May 27, 2019

Challenges in the simulation and analysis of turbulent flow

EEMCS building 36 • Mekelweg 4 • 2628 CD Delft
van der Poelzaal LB 01.220 • 16:00-17:00

We have invited <link 3me/afdelingen/process-energy/people/fluid-mechanics/drir-gerrit-e-elsinga/>Dr Gerrit Elsinga</link> from the Fluid Mechanics Chair at the Faculty of Mechanical Engineering next door to illustrate his perspective on the challenges arising from the direct numerical simulations (DNS) of turbulence. These are simulations of turbulent motions obtained by solving the governing Navier-Stokes equations without the a priori assumptions of ‘closure models’ that, in contrast, simplify the role of the most intricate spatial and temporal features considerably.

THE TALK

Flow turbulence is amenable to direct numerical simulation since the governing equations are known. These simulations, however, remain fundamentally challenging because of the wide range of spatial and temporal scales that need to be resolved, especially at the high Reynolds numbers of practical interest. Nevertheless, considerable progress in this area has also caused a partial shift of attention to the data analysis, which can be equally computationally demanding.

This talk discusses the analysis of turbulent motions from existing DNS. The understanding of these motions is important for predicting, for example, turbulent mixing and pollutant production in combustion. The results obtained for homogeneous and isotropic turbulence reveal that the small-scale straining motions converge to a state independent of Reynolds numbers. However, new substructures are predicted to develop at even higher Reynolds numbers, which is important when extrapolating the available results to applications in aviation, atmospheric science and astrophysics. Furthermore, it will be shown that the state-of-the-art simulations of the turbulence near walls have not reached that converged state yet.

BASIS FOR DISCUSSION

Accurate numerical treatments and intense compute workload ensue from resolving flow features down to the smallest temporal and spatial scales of fluid motion. Additional challenges lurk in the post-processing of the datasets and in extracting appropriate interpretations from them. We invite all guests to contribute actively to this event with their full range of expertise. The Group of Numerical Analysis is pleased to create an occasion to exchange ideas and intuitions fruitfully, and to envision opportunities for participating in research activities jointly.