Databases to assess the hydraulic conductivity of road construction materials

Geotechnical databases are useful for providing quantitative assessments of the variability of important geotechnical properties. Such databases are vital for the development and calibration of transformation models which can be used to estimate more complex variables from more fundamental parameters. This study examines two recently compiled databases of hydraulic conductivity of materials important to road construction, namely FG/KSAT-1358 for fine-grained soils, and AC/k-1624 for asphalt concrete mixtures. The influence of statistically classified outliers on the regression coefficients in recently published transformation models derived from these two databases is studied. It is shown that the regression coefficients from the transformation model produced using AC/k-1624 are more influenced by the outlier removal process compared with those from FG/KSAT-1358. The Akaike Information Criterion is used to evaluate the best fit probability density functions to the key database parameters. It is shown that the ‘loglogistic’ distribution is the best pdf to describe the variation of key parameters from both databases while the ‘lognormal’ distribution describes best the variation of the void ratio (e) data in FG/KSAT-1358.