Microfabrics-based Approach To Estimate Compressive Strength Using Multiple Regression And Artificial Intelligence Techniques

Uniaxial compressive strength (UCS) is one of the most important intact rock parameters, which commonly used for a variety of engineering designing surface and underground rock structures. Occasionally, in complex inherent characteristics of certain rocks, especially anisotropic rock it may be difficult to measure the UCS. Consequently, the use of empirical relationships to estimate UCS of rocks can be more practical and economical. Although, microfabrics properties are neglected during initial estimate of rock properties, they have great significance for the failure of rocks especially when intense deformation is present. Therefore, this study aimed to highlight the significance of rock microfabrics properties for predicting UCS of the intact rock in the absence of adequate budget and facility of testing UCS. This effort is a complementary contribution to the study presented herein.In this study, the prediction capability of artificial neural networks (ANNs) and fuzzy inference system (FIS) as well as the multivariate regression (MR) have been carried out to predict UCS from microfabrics properties of anisotropic amphibolite rocks. For this purpose, representative block samples of banded amphibolite rocks were collected from varied rock slopes along Wenchuan-Markang highway in Sichuan province, China. These rocks were selected due to their high rock-fabric complexity and their high significance impact in rock engineering design. In this study, the method of semi-automatic petrographic image analysis (PIA) was adopted to calculate and measure the microfabrics parameters. Data obtained from quantitative petrographic analysis and rock strength parameter was subjected to a series of comprehensive statistical analyses using SPSS V.19.0. Based on comprehensive statistical analysis, grain size, shape factor and quartz content were recorded as main microfabrics parameters affecting UCS values more than the other do. These three parameters have been adopted in this study to establish UCS models. To assess models performances some performance indices such as correlation coefficient (R), variance account for (VAF) and root mean square error (RMSE) were calculated and compared. The results revealed that even though the developed three models are reliable to predict the UCS, the study clearly indicates the superiority of the ANNs model. One of the most significant finding to emerge from this study is that the microfabrics-based PIA approach can easily extended to the modeling of rock strength in the absence of adequate budget and facility of testing UCS. Despite this result, the major disadvantages of neural networks (ANNs) compared to two models is their opaqueness of the underlying relationship between inputs and outputs.Furthermore, this study also assesses the strength and deformation anisotropy behavior of banded amphibolite rocks. Hence, the determination of anisotropic strength behavior of these rocks is essential for design purposes. The dynamic mechanical tests were performed on drilled rock samples as a function of foliation plane angle (β=0°,30°,60° and 90°). The results obtained have shown that the dynamic mechanical properties of amphibolite rocks have different values concerning banding plane. Compression and shear waves taken parallel to the foliation plane show highest values than those obtained in the other directions. It seems that the high range value (0.96-1.47) of strength anisotropic index is mainly due to slight undulation of foliation planes, that being not perfectly straight, whereas elastic deformation test show that there is no clear dependence on microstructures characteristics of subtype-amphibolite rocks that controlling modulus "shape-anisotropy". However, Young modulus values of amphibolite rocks with β follow both types of shape anisotropy, "U-shape" and "decreased order-shaped". Thus, this study recommended further research could be undertaken regarding the role of modulus "shape-anisotropy" within the same litho-type.