Quick and reliable information about the engineering properties of soils is needed for evaluating soils for nonagricultural uses. The objectives of this preliminary study were, (i) to relate some engineering properties of typical tropical soils from Nigeria to their mineralogy, (ii) to develop simple models for predicting and characterizing these engineering properties from easily determinable soil properties, and (iii) to compare values of the swell potential of these soils estimated by various equations used in soil mechanics with values measured directly in the laboratory. Data from 30 soil samples collected across Nigeria were used for this study. The extent of the volumetric shrinkage (VS) hazard posed by the soils is influenced more by the dominant clay minerals than by the texture or soil order. Generally, soils containing only kaolinite (1:1 mineral) had low VS values (<10%), whereas those containing either low or trace concentrations of smectite or vermiculite (2:1 minerals) acted like expansive soils (VS = 10-30%) in spite of the dominance of the 1:1 mineral. Soils with substantial amounts of either smectite or vermiculite had very high VS values (>30%). There were strong linear relationships between VS, liquid limit (LL), and plastic limit (PL) on one hand and cation exchange capacity (CEC), organic matter (OM), and clay content on the other. Regression equations developed with CEC could not predict accurately the VS and LL of two of the 10 validation soils with <5 cmol(c)/kg CEC and dominated by kaolinite. The OM-based model could not provide good estimates of the PL of the two vertisols with >80% VS values. The relationships between the soil properties and plasticity index were weak. Some nonlinear equations used to predict the swell potential of soils grossly underestimated the measured shrink-swell hazard posed by these soils. These predictive empirical equations need to be tested on a wider range of tropical soils than were used in this study before any valid conclusions on their applicability can be made.
Soil Science 01/1998; 163(2):93-102. DOI:10.1097/00010694-199802000-00002