Regression-Based Prediction of Soil Plasticity and Consistency Indices in the Geological Context of the Sibiu Area, Romania

Atterberg-limit testing is fundamental to geotechnical characterization, although the number of determinations may be limited in routine infrastructure investigations by the available testing programme. This study applies and validates previously established regional regression relationships for predicting the plasticity index (I_p) and consistency index (I_c) of fine-grained soils from the Sibiu area, Romania. A database of 83 samples with complete grain-size fractions, natural water content (w), liquid and plastic limits (w_L, w_P), depth (z), and Atterberg indices was analysed. The plasticity index was predicted from the liquid limit through the linear model I_p = 1.06627 w_L − 17.4372, according to [1], whereas the consistency index was predicted using a logarithmically transformed multivariable regression incorporating depth, clay fraction, and natural water content. The I_p model achieved R² = 0.912 and RMSE = 3.64%. The transformed I_c regression achieved R² = 0.942. Lower and upper characteristic values for Ip and I_c were derived from the regression predictions using the statistical procedure described in the manuscript. The proposed relationships support preliminary site characterization, interpolation between tested horizons, and quality control of laboratory determinations; they complement rather than replace laboratory testing. Their applicability is limited to geological conditions comparable with the investigated fine-grained Neogene deposits of the Sibiu area.