Analysis of soil compression curves from uniaxial confined compression tests

The compressive behaviour of soil, the compression curve, is typically expressed in a semi-logarithmic diagram in the relationship between the logarithm of applied stress, σ, and the void ratio, e. The compression curve yields three important soil compression properties: the swelling index, Cs, the compression index, Cc, and the precompression stress, σpc. The compression index is considered an indicator of soil resistance to compaction, Cs is used as a measure of soil mechanical resilience, and σpc is considered the soil load support capacity. The objective of this paper was to investigate the impact of soil particle and aggregate size distribution on the characteristics of the soil compression curve and on Cs, Cc, and σpc for a range of Swedish soils. We found no or only weak impacts of particle size and aggregate size distribution on soil compression properties. Instead, soil compression properties were largely controlled by initial void ratio, e0, in such a way that Cc and Cs decreases, and σpc increases with decreasing e0, although the correlation between σpc and e0 was not statistically significant for subsoil samples. We show that the soil compression properties (σPC, Cc and Cs) and their dependency upon e0 are partly a consequence of the compression curve being analyzed in a semi-logarithmic (log σ − e) diagram. The use of log σ (instead of σ) forces the compression curve to bend at a certain value of log σ, and this bend is associated with σpc. We showed that a precompression stress, as commonly defined, can be obtained for an ideal linear-elastic material. Therefore, the use of a logarithmic stress scale may lead to misinterpretation of soil mechanical behaviour. Our empirical results and theoretical exercise emphasize the need for further studies of physically-based expressions for soil strength for use in compaction modelling.