With exascale computing expected to arrive relatively soon, the development of new parallel algorithms that successfully utilize next-generation extreme-scale architectures is receiving much attention in many different domains of science.  In the case of intermittency in fluid turbulence, production results in isotropic turbulence at high Reynolds number and at grid resolutions as high as 18432-cubed are now available from a GPU code designed to allow the largest problem size feasible.  However, since resources available are finite, simulations at the largest problem size typically tend to be short, and their ability to provide sufficiently many independent realizations  for high-quality ensemble-averaging is in doubt.  In this talk we consider a new paradigm, in which simulations at the largest problem size are used as a tool to improve upon the by accuracy of physical information suggested by less-expensive simulations at lower resolution in time and space, but nominally the same physical parameters (i.e., the Reynolds number). Simulation datasets at lower resolution are either available from previous work or relatively inexpensive to run anew for a period of several large-eddy time scales. We take multiple snapshots from such a lower resolution calculation as initial conditions for simulations at higher resolution. The degree of independence between such ``initial'' snapshots can be quantified by a coherency spectrum and conditional statistics been dissipation rate fluctuations computed from each snapshot. As one might suspect, perhaps based on Kolmogorov's hypotheses,  the small scales are found to evolve quickly in response to an increase in resolution, which means simulations at the large problem size do not need to be long.  This approach allows a careful accounting of errors due to resolution in time, resolution and space, and statistical sampling, which maximizing the value of hero-sized simulations performed only over a short time span.


The use of extreme-scale turbulence simulations for studies of fine-scale intermittency

The Virtual Technical Meeting of the Society of Engineering Science 2020 is the first fully virtual meeting of the SES community to envision and connect during the hiatus created by Covid-19, providing a platform for the exchange of ideas, and to showcase the latest research in the field of engineering science.

[Visit the Programme](https://drive.google.com/file/d/1BoCPBzrr1IQFANuzQI8a-paXCJoOcCkB/view?usp=sharing) | [Visit the Prerecorded Video/Poster List](https://drive.google.com/file/d/1XHLeEXJxEFjCSsNScuLDfMwy1X7oy7oJ/view?usp=sharing) | [View the Session Grid](https://drive.google.com/file/d/1VeDOC-Ev5m_ymYjXnjaXSZHR3_DARiLt/view?usp=sharing)