Modifying Young's modulus in DEM simulations based on distributions of experimental measurements

Abstract The discrete element method, as currently employed by members of the fusion community, is rooted on the assumption that each pebble is a perfectly elastic material that obeys Hertz's theory for normal interaction. This assumption impacts the magnitude of inter-particle forces predicted by the models. We scrutinize the Hertzian assumption with single-pebble crush experiments with carefully recorded force-displacement responses and compare them to the non-linear forces predicted by a Hertzian pebble with bulk properties reported in literature. We found each pebble generally has a non-linear force response but with varying levels of stiffness that qualitatively matched the curves from Hertz theory. Assuming Hertzian interaction, we backed-out an elastic modulus for each pebble. We define a softening coefficient, κ, as the ratio of the pebble's elastic modulus to the sintered bulk value from literature. After determining the κ value for every pebble in our batch, we discovered a probability distribution for different batches. The distribution is attributed to the varying micro-structure of each pebble. We incorporate the results into our DEM algorithms, distributing κ values at random to pebbles satisfying the probability curves of experiments. DEM simulations of pebble beds in oedometric compression are carried out to determine macroscopic responses of stress–strain, contact force distributions at maximum stress, and a prediction of pebbles crushing at that point. In all cases studied here, the pebble beds with modified Young's modulus had smaller overall contact forces and fewer predicted crushed pebbles.