Date of Award
1-1-2003
Document Type
Thesis
Publisher
Edith Cowan University
Degree Name
Master of Science
Faculty
Faculty of Communications, Health and Science
First Supervisor
Associate Professor Lyn Bloom
Second Supervisor
Dr Ute Mueller
Abstract
Nickel and cobalt are key additives to metal alloys in modem industry. The largest worldwide nickel-cobalt resources occur in nickel laterite deposits that have formed during the chemical weathering of ultramafic rocks at the Earth's surface. At the Murrin Murrin mine in Western Australia, the nickel laterite deposits occur as laterally extensive, undulating blankets of mineralisation with strong vertical anisotropy, near normal nickel distributions, and positively skewed cobalt distributions. The mineral resources in nickel laterite deposits in Murrin Murrin are usually estimated from drilling and sampling on relatively wide-spaced drill patterns that are supported by local clusters of close-spaced sampling. The combination of deposit geometry and sampling configuration presents several estimation challenges for geostatistical resource estimation methods. In this thesis, close-spaced grade control drill sampling data from Murrin Murrin is used to quantify the estimation effectiveness of the wider spaced actual exploration pattern used to define the original resource, and an alternative cost saving stratified sampling pattern. Additionally, an unfolding of the laterite blanket by vertical data translation prior to nickel and cobalt grade estimation is tested for each exploration pattern. The unfolding essentially removes undulations in the laterite blanket prior to grade estimation by vertical translation of the sample data relative to a surface of high grade nickel-cobalt connectivity. Unfolding is expected to improve estimation accuracy in terms of grade and volume, as well as improve the quality of variography analyses. The stratified pattern is expected to give similar estimation accuracy to the actual exploration pattern. The effectiveness of ordinary kriging and full indicator kriging estimation algorithms from GSLIB software are compared for the combinations of in situ and unfolded cases of the actual sampling pattern used to define the deposit and an alternative stratified sampling pattern. For each combination, the estimates are made at the data locations of closed spaced grade control ‘reality'. The accuracy of each estimate is quantified by comparing the error, degree of bias and pseudo grade-volume relationships of the estimate to the 'reality' data. Additionally, the quality of exploration pattern variography is assessed against the grade continuity of the grade control information. Importantly, the main focus of these comparisons is on the correct estimation of local high grade nickel and cobalt resources that are preferentially processed in the early years mining. The results of comparisons between estimation methods and sample configuration combinations investigated show that the combination of unfolding and indicator kriging gives the best correspondence (in terms of grade and volume) of the various estimates to the grade control reality. The results of comparisons between the actual and the alternative stratified exploration pattern show that the cost saving' alternative pattern produces estimates similar to the actual exploration estimates.
Recommended Citation
Murphy, M. P. (2003). Geostatistical optimisation of sampling and estimation in a nickel laterite deposit. Edith Cowan University. Retrieved from https://ro.ecu.edu.au/theses/1295