Application of multivariate regression analysis (MLR) for predicting the UCS and E of sandstones using petrographic characteristics

The accurate determine of uniaxial compressive strength (UCS) and modulus of elasticity (E) using laboratory methods requires substantial time and cost. To overcome these difficult, development of predictive empirical equations to estimate the UCS and E of rocks is very important in rock engineering. This study deals with the prediction of UCS and E of sandstones from petrographic characteristics using multivariable linear regression analysis (MLR). For this purpose, 20 rock blocks were collected from sandstones in different locations of Upper Red Formation in southwestern Qom. Samples were subjected to petrographic examination, which included the observation of 16 parameters and modal analysis. The specimens were also tested to determine the uniaxial compressive strength, modulus of elasticity, porosity, and dry density. Based on the results of statistical analysis, multiple predictive equations were developed to estimate the mechanical properties using petrographic characteristics. The performance of developed equations was assessed using R, RMSE and VAF. Based on the results, it was observed that the proposed equations have a good performance in predicting the UCS and E.