Project Details



The aim of this project is to reduce the computational complexity of the Discrete Element Method (DEM). Here, we plan to use Machine Learning to train and build a more reliable model for fluid forces on individual particles that are part of agglomerates. Even though Direct Numerical Simulation (DNS) provides full details related to forces acting on agglomerates, it is too computationally expensive to perform large-scale simulation using Direct Numerical Simulation (DNS). Whereas the Discrete Element Method is widely used & comparatively easy to perform simulations for particle counts of O (106 to 107). Later, we plan to combine this framework into the Discrete Element Method (DEM) to improve forces calculation which otherwise uses various empirical correlations which are not accurate when compared to Direct Numerical Simulations. Once we achieve a certain level of surety, we want to implement this model to improve the understanding of particle agglomeration and breakage phenomena.
This project is thus concerned with improving the dynamics of forces and flow (focusing on Drag) of the Discrete Element Method applied to particulate flow using supervised Machine Learning from higher fidelity data set obtained from Direct Numerical Simulation. Using this trained Machined Learning model, we could accurately predict dynamics driving the flow, particle-particle, and particle-flow interaction (i.e., contacts, adhesion, breakage) which otherwise are based on empirical relations equations and are averaged values.

Funding:  Self-funded
Effective start/end date01/08/202131/07/2024


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.