AI meets particle expertise to simplify flowability and packing density predictions

Round particles and their properties are simple to explain mathematically. But the much less spherical or spherical the form, the tougher it turns into to make predictions about their conduct. In his doctoral thesis on the Technical University of Kaiserslautern (TUK), Robert Hesse has skilled a neural community to routinely decide the packing density and flowability of non-spherical particles.

Few particles in nature or in industrial production are precisely spherical; as a substitute, there are a mess of variants and shape traits. This is precisely what makes it so sophisticated to explain non-spherical particles and optimize their dealing with primarily based on the outline. For instance, the rounder a pill is, the much less seemingly it’s to snag on different tablets within the filling course of. A flat cylindrical form can already be optimized by slight rounding on the subject of packing density.

But how can all of the properties that decide flowability and packing density be shortly recorded in an effort to derive choices on the selection of a form? What beforehand required simplified calculations of particular person mathematical parameters or mould elements might be derived routinely by a skilled synthetic intelligence—on this case a so-called “Deep Convolutional Neural Network”—utilizing a 3D mannequin.

“Using simulations in which only the shape of the particles varied, I created a comprehensive experimental data set and used it to train the neural network,” stories Hesse, a analysis affiliate on the Department of Mechanical Process Engineering. “Standardized experiments with 3D-printed particles allowed the simulation methodology to be validated in the test phase—that is, to match how accurately the simulation can represent real particles.”

The skilled neural network now filters out salient options corresponding to curves, corners, edges, and so forth. from any three-dimensional level cloud representing the complete form. Using this data, it might analyze flowability and random packing density. “This is useful, for example, for optimizing the shape of pharmaceutical products in terms of minimum machine dimensions and package sizes,” the researcher says.

Provided by
Technische Universität Kaiserslautern