Poster-CSSI-210387-Meneveau-2024

This project deals with stewardship and democratization of big data from computational fluid turbulence numerical simulations. We perform high-fidelity simulations and instead of analyzing data on the fly and then having to throw data out we store entire dataset --- spatio-temporal fields --- in a database. We build user-friendly interfaces that enable users to place virtual sensors in the flow and extract only regions of interest. This approach has already proven to vastly enhance access to what otherwise are difficult to share datasets. This Framework project is focused on transitioning from a SQL-server based system to a more scalable and robust workflow. Analysis tools based on Jupyter notebooks enable more user-programmable functions close to the data. The project adds new datasets of interest to the geosciences and fluids engineering communities, growing the system from half a petabyte now to 2 petabytes when completed. This year we report on the completed new dataset for stably stratified atmospheric boundary layer and progress on the (32k)-cubed isotropic turbulence simulation. We will provide an update on the current progress and experiences developing a new python notebook and Matlab code which with a REST interface enables flexible access to the data. Some science outputs using the data will also be highlighted, such as a new thermodynamic formalism of turbulence and study of the Kolmogorov Refined Similarity Hypothesis.