PREDICTION OF ENGINEERING PROPERTIES OF A SELECTED LITHARENITE SANDSTONE FROM ITS PETROGRAPHIC CHARACTERISTICS USING CORRELATION AND MULTIVARIATE STATISTICAL TECHNIQUES

The determination of rock strength and elastic parameters in laborator y tests requires high quality core samples which are not always availa ble from fractured and/or weathered strata. The purpose of this study is to apply a correlation analysis in conjunction with multivariate st atistical techniques to petrographical and laboratory geomechanical da ta for the investigation of correlations that exist between these prop erties of a selected sandstone, and to suggest inexpensive predictive models in a preliminary investigation. Sandstone samples from a boreho le drilled for a preliminary investigation of a shaft location at a ha rd coal region (Black Sea, Turkey) were tested to determine uniaxial c ompressive strength, elastic parameters, unit weight, porosity, point load index and quality index. Secondly, fifteen selected samples were subjected to petrographic examination which included fifteen parameter s and modal analysis. Based on the statistical analysis results, polyn omial prediction equations were developed for estimating mechanical pr operties from petrographic characteristics and index properties. The s tudy revealed that the influence of textural characteristics appear to be more important than mineralogy for predicting engineering properti es. It is also noted that type of contacts, grain shape and size, rock fragments with packing density and proximity are the petrographic cha racteristics which have significant influence upon engineering propert ies of the sandstones. Results of this study were presented in the for m of correlation coefficient models and predictive models.