Use of Machine Learning for unraveling hidden correlations between Particle Size Distributions and the Mechanical Behavior of Granular Materials
This work addresses a domain-specific problem in geotechnical engineering by providing a data-driven method to uncover hidden correlations, though it is incremental as it applies existing ML techniques to a new dataset.
The study tackled predicting the mechanical behavior of granular materials from particle size distributions using a neural network trained on DEM simulations, achieving considerable accuracy despite noisy data.
A data-driven framework was used to predict the macroscopic mechanical behavior of dense packings of polydisperse granular materials. The Discrete Element Method, DEM, was used to generate 92,378 sphere packings that covered many different kinds of particle size distributions, PSD, lying within 2 particle sizes. These packings were subjected to triaxial compression and the corresponding stress-strain curves were fitted to Duncan-Chang hyperbolic models. A multivariate statistical analysis was unsuccessful to relate the model parameters with common geotechnical and statistical descriptors derived from the PSD. In contrast, an artificial Neural Network (NN) scheme, trained with a few hundred DEM simulations, was able to anticipate the value of the model parameters for all these PSDs, with considerable accuracy. This was achieved in spite of the presence of noise in the training data. The NN revealed the existence of hidden correlations between PSD of granular materials and their macroscopic mechanical behavior.