Improving simulations of large-scale dense particle-laden flows with ma- chine learning: a genetic programming approach

Particle-laden flows are encountered in many natural and industrial processes, such as, for instance, the flow of red and white blood cells in plasma, or the fluidization of biomass particles in furnaces. Over the last 40 years, scientists have used Euler-Lagrange (EL) simulations as a way to predict the behavior of such flows. However, EL simulations rely on models to describe the interaction between the fluid and the individually tracked particles. These models require the so-called "undisturbed” fluid velocity at the location of the particle, which is what the velocity of the fluid would have been if the particle had not been there. Current models for this are very rudimentary and precisely calculating the undisturbed fluid velocity is extremely expensive, as it would involve running many additional highly resolved simulations of the same case where one particle is left out.

This is a project to deliver a novel model for the undisturbed fluid velocity at each particle location, given the properties of the flow around the particle and of the surrounding particles, using a supervised learning machine learning approach: genetic programming (GP). GP is highly suitable, as its result will not be a "black-box” model, but a verifiable expression for the undisturbed velocity. This expression will be validated by analytical solutions and highly resolved simulations, and will enable accurate, large-scale simulations of dense particle-laden flows, while only requiring a fraction of the cost of fully resolved simulations.