Department Seminar of - Liron Saimon - Extending understanding of the problem of solid particles settling through a density interface using ML

30 November 2022, 14:00 - 15:00 
פקולטה להנדסה 
0
Department Seminar of - Liron Saimon - Extending understanding of the problem of solid particles settling through a density interface using ML

School of Mechanical Engineering Seminar
Wednesday, November 30, 2022 at 15:00
Wolfson Building of Mechanical Engineering, Room 206

Extending understanding of the problem of solid particles
settling through a density interface using ML

By Liron Simon Keren

M.Sc. student of Prof. Alex Liberzon

Turbulence Structure Laboratory

 

Settling due to gravitational force or flotation due to buoyancy are basic transport mechanisms of solid bodies in fluids on Earth, such as in marine snow sedimentation, CO2 capture in stratified lakes, and flotation processes in industrial applications. In nature and industry, fluids are often in-homogeneous, where dissolved substances and temperature differences act as density-stratifying agents. Objects crossing these regions of strong density gradients have been observed to experience an increased resistance compared to non-stratified fluid layers.

Experimental studies attempting to measure the additional resistance force on small settling particles across density interfaces are very challenging and require refractive index matching, careful optical setups and three-dimensional Lagrangian tracking with high spatial resolution. Thus, they result in a small number of trajectories and a limited set of parameters.

This study presents a different approach – we simplify experiments and measure the outcome of the physical process - the retention time, which is the extended time a particle takes to cross and reach the terminal velocity in the bottom, denser fluid layer. This approach allows us to significantly extend the parameter space into previously unexplored ranges and provides sufficient data for Machine Learning methods. Furthermore, we developed a new ML system for symbolic regression, explicitly designed for physics-informed problems. Using this ML system, we produce two predictive correlations of retention time for an extensive range of particle and fluid parameters: a) in the form of a trained ML, and b) in the form of a symbolic equation.

Retention time correlations in an extended parameter range can help estimate the effect of increased residence times on CO2 capture or microplastic settling and improve the effectiveness of industrial processes with stratified fluids for better sustainability.

 

 

 

Join Zoom Meeting

https://tau-ac-il.zoom.us/j/4962025174?pwd=bVJUeElXRUUya3BERisyNllLOE9EZz09

 

 

 

 

 

Tel Aviv University makes every effort to respect copyright. If you own copyright to the content contained
here and / or the use of such content is in your opinion infringing, Contact us as soon as possible >>