Offering models for estimation of compressibility coefficient in fine grain soils

The compression index of soil masses is one of the important soil parameters that is essential for geotechnical designs. As the determination of the compression index from consolidation tests is relatively time-consuming, empirical formulas based on soil parameters can be useful. Over the decades, a number of empirical formulas have been proposed to relate the compressibility to other soil parameters, such as the natural water content, liquid limit, plasticity index, specific gravity, and others. In this study, first by simple and multi linear regression based on 115 case study, empirical formulas proposed to relate the compressibility to other soil parameters, such as the, liquid limit, plastic limit, plasticity index, void ratio at liquid limit, void ratio at plastic limit. An alternative approach, an artificial neural network (ANN) model, is proposed to estimate the compression index with numerous consolidation test sets. The compression index was modeled as a function of five variables including the plastic index, clay fraction, void ratio at liquid limit, void ratio at plastic limit and specific gravity. Finally, comparison between proposed models carried out. The ANN model has a significantly better performance than the empirical equations for the soil compression index.