The correlation between RMR89 and GSI on tunnel 3 Sigli Banda Aceh toll road

• Main Authors: Latif, Hanif Khoirul, Husein, Salahuddin, Setiawan, Hendy
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; Boreholes; Classification; Construction sites; Correlation analysis; Regression models; Rock mass rating; Root-mean-square errors; Toll roads; Tunnel construction; Tunnels
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0152409

Tunnel 3 of Sigli Banda Aceh toll road in Indonesia is currently in the planning stage. The rock mass quality at Tunnel 3 has been determined using RMR method based on subsurface rock samples from 4 boreholes. However, there is a high degree of uncertainty that resulted from this rock mass quality classification. Therefore, it is necessary to do correlation analysis for several classification methods when determining the rock mass quality. Another classification method that often be used in determining rock mass quality, especially in road tunnel projects, is Geological Strength Index (GSI). This study aims to review the correlation between RMR89 and GSI values at Tunnel 3 construction site, which based on four drill core data containing 186 samples. The method used in this study is by classifying the rock mass quality with both RMR89 and GSI methods and generating their correlation through the trend lines. The trend lines then applied to formulate regression models equations. The regression models then were tested for their representativeness level and followed by the evaluation of each precision and accuration. Precision examination is carried out by calculating the value of Pearson’s coefficient for correlation (R), mean absolute error (MAE), and root mean square error (RMSE) from each regression models. Meanwhile, the accuracy examination is carried out by computing the mean absolute percentage error (MAPE). From this research, it can be stated that the correlation of RMR89 and GSI can be presented in five regression models. Each model has a satisfying level of determination coefficient. The regression models also have a high degree of precision with satisfying R, RMSE, and MAE. MAPE calculations from the proposed models also show a high degree of accuracy. From the degree of precision and accuracy, it can be concluded that five generated regression models of RMR89 and GSI correlation are quite suitable to be applied into research area. The application of these mathematical models at the research are expected to provide additional considerations in classifying rock masses which can result in increased accuracy for determining excavation methods and tunnel stability.