Comparison of Mathematical Models for the layout of Granulometric Curves of Brazilian Soils

Abstract

The particle size distribution of solid particles (PSD) is critical in determining the potential for compaction, the availability and the redistribution of water in the soil, especially in the areas of building material and soil mechanics, among others. However, many banks of soil data does not contain detailed PSD data, but only some fractions. A mathematical accurate representation of the PSD is required to estimate soil hydraulic properties and texture comparing measurements of different classification systems. The objective of this study was to compare the performance of 12 models, with 2 and 3 tuning parameters proposed in the literature to represent the PSD, and, predict the water retention curve in the soil, from a wide range of Brazilian soils textures. The statistical parameters (NDEs, RD, MS CRM) showed three models tuning parameters proposed by Lima & Silva, Weibull and Fredlund got the best performance, with lower NDE, RD, MS very close to one and CRM values very low. The models Lima & Silva, Weibul, Fredlund and Skaggs, with three tuning parameters, and the models Skaggs end Lima & Silva, with two parameters, proved to be suitable for estimate the water retention curve in the soil, for soils with coarse and fine texture. Keywords: Granulometry; Soil texture; Curve Adjustments.