Estudo da resistência ao cisalhamento de solos via regressão linear

Abstract

Soil shear strength refers to the soil's ability to resist tangential forces acting on its shear planes. This geotechnical property is essential in analyzing the stability of slopes, foundations, embankments, among other geotechnical structures. This work aimed to develop a predictive linear regression model for the shear strength of soils as a function of some characteristics of the analyzed soil and the measurement method. The data used comes from a mining area in the interior of the state of Minas Gerais, Brazil, observed between 2017 and 2023. A model with high predictive power (R²=0.982) was obtained, containing information on five variables: test type, p 'rupture, variation between the initial and final void ratios, initial moisture and fines content. The signs of the estimated coefficients make sense in practice, indicating that, controlled by the other factors, the average shear strength increases as rupture and fines content increase, but reduces as the initial soil moisture increases and if the test is not - drained. A more in-depth analysis of residuals and the application of correction methods can contribute to improving the reliability of inferences, for example, by revisiting possible anomalous observations and/or leverage points. In any case, the analyzes carried out can contribute to the improvement of Geotechnical Engineering practices, especially in activities related to mining, offering tools for evaluating soil stability.