Random close packing fractions of lognormal distributions of hard spheres

TL;DR

This paper demonstrates a fast one-dimensional algorithm for predicting the packing densities of polydisperse spheres with lognormal size distributions, matching results from slower 3D simulations and enabling inverse design of particle packings.

Contribution

The authors adapt a 1D algorithm to efficiently predict packing fractions for lognormal distributions, facilitating inverse design and comparison with 3D simulation methods.

Findings

Algorithm accurately predicts packing fractions.

Results compare well with 3D simulation algorithms.

Enables inverse problem solving for particle size distributions.

Abstract

We apply a recent one-dimensional algorithm for predicting random close packing fractions of polydisperse hard spheres [Farr and Groot, J. Chem. Phys. 133, 244104 (2009)] to the case of lognormal distributions of sphere sizes and mixtures of such populations. We show that the results compare well to two much slower algorithms for directly simulating spheres in three dimensions, and show that the algorithm is fast enough to tackle inverse problems in particle packing: designing size distributions to meet required criteria. The one-dimensional method used in this paper is implemented as a computer code in the C programming language, available at http://sourceforge.net/projects/spherepack1d/ under the terms of the GNU general public licence (version 2).