A Statistical Study Of Relationships Between Rock Properties

A knowledge of rock characteristics forms the foundation for the application of rock mechanics to engineering works. Hence, there has been an increased interest in the determination and proper utilization of rock properties data. Most numerical values quoted for rock properties are obtained from arbitrary tests and are not necessarily indicative of any intrinsic rock characteristics. Recent attempts to correlate these properties by statistical methods have attracted considerable attention, in the hope that these might give some clue to the fundamental properties. To date, correlation and regression techniques have been employed for this purpose. The use of statistics has also been deemed appropriate because of inhomogeneity and anisotropy in rocks. The problems have been further compounded by the lack of standardized methods of measuring rock properties. The various ASTM committees are still deliberating on these matters, but recommendations should be forthcoming in the near future. As more reliable data become available, the ability of geoscientists to find more meaningful relationships between rock properties will be greatly enhanced. The objective of this paper is to outline another statistical technique which might be useful in providing an insight into the relationships that exist between rock properties. This method does not attempt to predict rock characteristics, but rather compares the similarity or dissimilarity between properties. This enables rock properties to be classified into groups with similar intrinsic characteristics. The technique described is one form of factor analysis. It allows the results obtained to be depicted geometrically, so as to make ready visual examination possible.