The SCRAMBLE project has already demonstrated computationally that turbulence can be achieved at microconfined conditions. In particular, it has been discovered that turbulence can be sustained at significantly lower Reynolds numbers that in traditional (low-pressure conditions) wall-bounded flows. One of the main reasons is the appearance of a baroclinic instability resulting from the interaction between pressure and density gradients as the fluid transverse the pseudo-boiling.
At present, the project is focusing on double-checking the findings experimentally. In this regard, an experiment has been designed, mounted and tested in a laboratory facility. In particular, the system uses CO2 as a working fluid (critical pressure Pc = 73.8 bar), and it is fabricated using metal, glass-glass, and silicon-Pyrex components. Moreover, the thermal part of the experiment is based on imposing a temperature difference between the top and bottom walls of the microchannel. The top wall works at a higher temperature than the critical one (Thot > Tc), and the bottom wall at a lower temperature (Tcold < Tc). In detail, (i) at the hot wall, a PI system driving a Peltier is employed to heat and control the top wall of the microchannel to the required temperature; (ii) at the cold wall, a passive system with an aluminum mass is designed and developed to maintain the temperature at room temperature. Therefore, the experiment is ready to be used for research tasks, and the project will concentrate on this task over the next months.
From a data science perspective, the project has developed a set of post-processing software tools to obtain deep insight from the computational simulations and laboratory experiments. In particular, a machine learning framework has been developed, Python scripts have been generated for processing the flow physics of wall-bounded flows presenting variable thermophysical properties, and a predictive model for confined turbulence is being developed based on dynamic mode decomposition (DMD) methods.
In terms of design-oriented validation-verification & uncertainty quantification, the project has develop a pioneering methodology to reduce the dimensionality of the problem of interest. In detail, the methodology is a data-driven augmentation of the traditional Buckingham Theorem in fluid mechanics based on active subspaces. This strategy will be very useful in the coming year to facilitate the design and optimization of the Turbulence-On-a-Chip concept.
