The past fifteen years have produced rapid advances in network modeling, most notably the replacement of previous-generation stick-and-ball networks with physically representative models of real materials. Unfortunately, technology to generate these networks has lagged behind the advances for flow modeling, which has slowed the use of network modeling techniques in practical engineering problems.

We have invested significant resources into this problem during the past two years, the result being a series of powerful network generation algorithms. The best one to use depends on the structure of the material and the type of data set that describes the material. Distinctions include granular versus amorphous materials, tomography data versus computer-generated materials, and whether the domain has complicated boundaries.

.