Date of Award

2011

Degree Type

Thesis

Degree Name

Master of Science

School

School of Engineering

Faculty

Faculty of Computing, Health and Science

First Advisor

Associate Professor Ute Mueller

Abstract

Stochastic simulation of rocktypes, or the geometry of the geology, is a major area of continuing research as earth scientists seek a better understanding of an orebody as a precursor to the assignment of continuous rock properties, allowing more economically appropriate decisions regarding mine planning. This thesis analyses the suitability of particular geostatistical rock type modelling algorithms when applied to the five rocktypes evident in drill hole data from the Big Bell gold mine near Cue, Western Australia. The background of the geostatistical theory is considered, in particular the concept of the random function model and the link between the categorical statistics determined from the drill hole data and the three models used for estimation and simulation. The commonly applied indicator kriging (IK) and sequential indicator simulation (SIS) algorithms are compared in a non-sedimentary gold deposit environment to the more computationally demanding and more complex plurigaussian simulation (PGS). Comparisons between the three models are made by examining global and regional rocktype (lithotype) proportions of the outputs of the models, both visually and empirically. The models are validated by considering the contacts which occur in reality between different lithotypes and the proportion of contacts which do not conform to this reality in each of the models. This „inadmissible contact‟ ratio measures the short range validity of the estimation and simulation techniques. Finally, cores taken from the output of the models are compared to the drill hole data in terms of transition proportions between the twenty five possible transitions for the five lithotypes. Inadmissible contacts were at a minimum with PGS, and the visual and empirical natures of the PGS output were closely linked to the reality of the drill hole data. Whilst each model produced similar 3D images, PGS was a realistic balance between the clustering effect produced by IK and the fine mosaic effect from SIS. The PGS output numerically outperformed the other two models in terms of admissible contacts and connectivity, most closely matching the drill hole data. All results indicate that, whilst demanding to implement, PGS produces the most adequate model of the study region.

Share

Paper Location

 
COinS