Abstract:
|
Oil sand geologists seek to understand the relationships between ore characteristics from different locations and their processability. Facies grouping is one effective way to a) simplify the different facies found in orebody and link to different properties; b) help bitumen grade modelling; c) help the identification of the watersands unit. In this study, a statistical cluster analysis approach was proposed and applied to Shell geology dataset for facies grouping. First, percentiles of the distribution of chemo-facies variables are used as the clustering variables, which are able to robustly capture the characteristics of the interested distributions. Second, the agglomerative hierarchical clustering method is applied for facies grouping with Squared Euclidian distance coupled with the centroid algorithm as the distance measure. Finally, the agglomeration distance plot is introduced to assist in identifying the number of clusters. Compared with Shell in-house existing method, the new approach can be more effectively and robustly in capturing the characteristics of facies distribution, therefore the results are more accurate.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.