Representing topology and geometry of 3D ore microstructures with GMaps


Representing topology and geometry of 3D ore microstructures with GMaps

Menzel, P.; Teichmann, J.; van den Boogaart, K. G.

The determination of optimal processing chains of specific and complex ores is one of the main tasks in mineral processing. For this purpose, we want to simulate the complete process chain from ore body to flotation to optimize for specific input materials. The key issue is the knowledge of the 3D ore microstructure as starting point for crushing into mineral particles.
Based on statistical information of 2D samples from MLA images (mineral liberation analysis), 3D microstructure models founded on random mosaics are fitted using stochastic geometry.
Then again, simulated MLA data sets can be created from this 3D microstructure model and compared to real data to optimize simulation.
To represent the geometry and the topology of the microstucture models, a data structure is needed to perform the following main tasks. Firstly, we need to build a comprehensive, topologically consistent cell structure. Secondly, we need easy access to all geometric features for all k-cells (e.g. volume, surface, contact faces). Furthermore, operations like the breakup along defined particle borders, the creation of profiles along planes and the dual graph have to be provided in an eficient way. We decided to use our own Java-based implementation of the Generalized Maps (GMaps) data structure that supports the requested operations.
The opportunity of the GMaps concept is, that it only uses one specific data type, called Dart, and several associations between incident Darts, called involutions, to describe the complete topology. Thus, it allows an efficient traversal through all topological elements of our microstructure models.
The simulated microstructure, represented by a GMap, can be used for a milling simulation and the resulting particle streams might undergo, e.g., a flotation simulation. The 3D microstructure models are the essential linkage between the analytical and the processing part of our mineral processing simulation.

Keywords: GMaps; microsructures; stochastic geometry; data structures

  • Lecture (Conference)
    18th Annual Conference IAMG2017, 02.-09.09.2017, Fremantle, Australia

Permalink: https://www.hzdr.de/publications/Publ-25053
Publ.-Id: 25053