82
reads

Chih-Che Chueh

2012-01-17
10:30:00 - 11:30:00

308 , Mathematics Research Center Building (ori. New Math. Bldg.)

In many different research areas including pharmaceutical science and fuel cell engineering, there is a great need to better understand how particles are packed when forming. A more detailed knowledge of the packing process would be beneficial in order to understand and optimize the microstructure of different kinds of porous media.

Conventionally, the packing particles constituting a micro porous structure are simply considered in a round or spherical shape. Although there are still some exciting applications of spherical random packing, many naturally or artificially occurring particulate substances are characterized by a non-spherical shape. For instance, the porous structure in PEM or Solid Oxide Fuel Cells is comprised of arbitrarily-shaped particles such as ionomers, carbon fibers etc. Therefore, if ones still use spherical random packing to form a porous structure for their micro catalyst modeling, it would be difficult to accurately predict the cell performance as the packing particles are actually of arbitrary shapes, or at least not of round shape. In other words, reconstructing non-spherical packing porous structures numerically would be a suitable choice to tackle this issue and an inevitable part for modeling PEMFC or SOFC micro-structure catalyst.

In this talk, our efforts/attempts to develop a generalized algorithm of non-spherical random packing will be presented. In our non-spherical random packing, a particle is defined by an assembly of component spheres to approximately represent shapes intended. In this algorithm, we first randomly distribute particles evenly in a domain. Then, overlaps between two adjacent particles produce a restoring force. The restoring force not only gives the particles a translational movement but also produce a moment about the center of mass of the particle. The moment results in a rotation about the center of mass or a reorientation of the particle. Overlaps are removed by iteratively moving each particle a small distance in a direction determined by the restoring force on the particle and re-orientating each particle under the action of its resultant moment. In addition, we also summarize several data structures such as linked list, heap list etc that are used in our packing algorithm to efficiently sort our particles in order to improve our computational efficiency. At the end, we will show some quantitative companions made by other algorithm developed by other groups.