Determination of Unconfined Compressive Strength of rock is important in Geotechnical Engineering. Although laboratory test is the most direct way of rock compressive strength estimation, but UCS determination in laboratory is problematic if the rock masses are weathered because obtaining proper core segments is difficult. Hence, using index testing as an alternative for UCS prediction is investigated by researchers. It is well established that Indirect/Brazilian Tensile Strength is related to UCS. In this paper, to develop a correlation between UCS and BTS, collected data of laboratory tests on dry limestone specimen including 20 Unconfined Compression Tests and 20 Brazilian Tests have been used. Then, to apply Artificial Neural Networks, a Radial Basis Network is developed to reach a relationship between BTS and UCS. Based on the low Mean Squared Error of the network, a new correlation is introduced for prediction of the UCS of limestone core samples from BTS data.
Unconfined Compressive Strength (UCS) of rock is considered as an essential parameter in analysis of geotechnical problems such as rock blasting and tunneling. Although laboratory test is the most reliable and direct method for estimating UCS, direct determination of UCS in laboratory is time-consuming and expensive. In addition, in direct method of UCS determination, having sufficient number of high quality rock samples is a prerequisite. However, it is not always possible to extract proper cores for sampling purpose in highly weathered rocks. Therefore, the use of various correlations for UCS prediction has been highlighted in the literatures. These correlations often relate other rock index parameters such as point load index, rebound number of Schmidt hammer and indirect tensile strength of the rock to UCS. Implementing such correlations is of interest, mainly due to the fact that rock index tests have the advantages of being relatively fast and economical. Brazilian Test (BT) is used for indirect determination of tensile strength of rock samples. It is established that Brazilian tensile strength is related to UCS. One of the most agreed correlations between UCS and indirect tensile strength or Brazilian Tensile Strength (BTS) of the rock is highlighted in the study by Sheorey (1997). According to his study, the compressive strength of the rock is approximately 10 times its tensile strength. Nevertheless, Sheorey's strength ratio variation is high (Cai 2006) and consequently cannot be generalized due to the fact that rock behavior varies from place to place and is site specific. This paper proposes a new correlation between UCS and BTS of specific type of rock i.e. limestone as the relationship between compressive and tensile strength of rock depends on rock type (Brook 1993).