Neural Tree for Estimating the Uniaxial Compressive Strength of Rock Materials

Abstract

Uniaxial Compressive Strength (UCS) is the most important parameter that quantifies the rock strength. However, determination of the UCS in laboratory is very expensive and time-consuming. Therefore, common index tests like point load (Is-50), ultrasonic velocity test (Vp), block punch index (BPI) test, rebound hardness (SRH) test, physical properties have been used to predict the UCS. The objective of this work is to develop a predictive model using a neural tree predictor that estimates the UCS with high accuracy and assess the effectiveness of different index tests in predicting the UCS of rock materials. UCS and indices such as BPI, Is-50, SRH, Vp, effective porosity and density were determined for the granite, schist, and sandstone. The constructed model predicted the UCS with a high accuracy and in a quick time (9 s). Additionally, the destructive mechanical rock indices BPI and Is-50 proved to be the best index tests to estimate the UCS.