Spatial modelling of geological domains with multiple training images: Application to the Red Dog mine, Alaska, United States

CIM Journal, Vol. 6, No. 3, 2015, pages 137-148

D. A. Silva
Centre for Computational Geostatistics, University of Alberta, Edmonton, Alberta, Canada

K. J. Palmer
Teck Resources Ltd., Vancouver, British Columbia, Canada

C. V. Deutsch
Centre for Computational Geostatistics, University of Alberta, Edmonton, Alberta, Canada

http://dx.doi.org/10.15834/cimj.2015.15
Abstract Evaluating resources and reserves requires the reliable modelling of geological domains and their grades. “Multiple point statistics” simulation is a relatively new geostatistical method to generate complex, high-resolution geological models by extracting high-order spatial relations from training images. A method based on combining multiple training images is developed to obtain the correct amount of large-scale continuity and short-scale variability. A calibration process matches the geological spatial variability between the final realizations and the drillhole dataset. A case study with data from the Red Dog mine, Alaska, United States, shows an appropriate reproduction of short-scale variability along boundaries between geological units.
Keywords: Entropy, Geological domains, Geostatistics, Linear opinion pool, Multiple point statistics (MPS), Training image (TI)
Résumé L’évaluation des ressources et des réserves exige une modélisation fiable des domaines géologiques et de leur teneur. La simulation par « statistique multipoint » est une méthode géostatistique relativement nouvelle pour générer des modèles géologiques complexes à haute résolution par l’extraction de relations spatiales d’ordre supérieur à partir d’images d’entraînement. Une méthode basée sur la combinaison de multiples images d’entraînement est développée afin d’obtenir la bonne quantité de continuité à grande échelle et de variabilité à petite échelle. Un processus d’étalonnage fait correspondre la variabilité géologique spatiale entre les réalisations finales et l’ensemble des données de forage. Une étude de cas avec des données de la mine Red Dog, Alaska, États-Unis, montre une reproduction adéquate de la variabilité à petite échelle le long des limites entre les unités géologiques.
Mots-Clé Combinaison linéaire d’avis différents, domaines géologiques, entropie, géostatistique, image d’entraînement, statistique multipoint.
Full Access to Technical Paper
PDF version for $20.00
Other papers in CIM Journal, Vol. 6, No. 3, 2015
Search
Sort By:  Relevance
Showing results 1 - 5
Text
Publication: CIM Journal
Issue: 3
Volume: 6
Year: 2015
Text
Summary: The degree of uncertainty associated with a natural diamond resource is important to quantify from the time of discovery through the production lifetime. Data collection occurs during the discovery, exploration, delineation, and production or recovery phases. Quantifying the relationship between data and uncertainty is an important component of project valuation. The value of data is measured as their potential to reduce uncertainty if they are collected. A method is developed using Monte...
Publication: CIM Journal
Author(s): J. G. Manchuk, J. Stiefenhofer and M. Thurston, C. V. Deutsch
Keywords: Diamond exploration, Geostatistics, Kimberlite, Monte Carlo simulation, Resource uncertainty
Issue: 3
Volume: 6
Year: 2015
Text
Summary: As part of a series describing the nickel operations of Vale Canada (formerly known as Inco), this paper focuses on the 1973 to current operations of the Copper Cliff nickel refinery in Ontario, Canada. The process used there employs the formation of Ni tetracarbonyl at high CO (gas) pressure and its subsequent decomposition at somewhat elevated temperatures to produce highly pure Ni pellets and powders. Personal sampling of workplace aerosol dusts is reported, with information on specific Ni...
Publication: CIM Journal
Author(s): B. R. Conard
Keywords: Aerosol dusts, Inco, Nickel refining, Nickel tetracarbonyl, Vapometallurgy
Issue: 3
Volume: 6
Year: 2015
Text
Summary: High-risk underwater mining operations necessitate characterizing spatial heterogeneity and resource uncertainty. Local average subdivision is a method of simulating spatial deposit variability inside a cell through a sequence of discretization stages, during which a cell is further subdivided into four or eight cells. New developments are presented to validate simulated deposit models and deal with point measurements of geotechnical and ore-grade properties. This paper illustrates the...
Publication: CIM Journal
Author(s): T. Wambeke and M. Alvarez Grima, G. A. Fenton, J. Benndorf, A. Vervoort
Keywords: Geostatistical simulations, Local average subdivision (LAS), Marine deposit evaluation, Resource uncertainty, Spatial variability, Underwater mining
Issue: 3
Volume: 6
Year: 2015
Text
Summary: Corrosion of support system components (e.g., bolts, mesh, plates) is a safety and economic concern in underground mines. This paper reports the results of a five-year study wherein corrosion coupons were installed at select locations in participating mines, recovered at regular intervals, analyzed for signs of corrosion, and tested to quantify loss in capacity. Complementary tests were conducted on recovered mesh and friction bolt specimens. The data provide insights into contributing...
Publication: CIM Journal
Author(s): J.-F. Dorion, J. Hadjigeorgiou, E. Ghali
Keywords: Corrosion, Reinforcement systems, Support systems, Underground mines
Issue: 3
Volume: 6
Year: 2015
Powered by Coveo Enterprise Search