Pavements and railway beds are badly affected by the behavior of their expansive clay subgrades. Seasonal drying and wetting are particularly responsible for irregularities in the pavement surfaces through differential settlements and heaving. In Israel, a method of quantifying the amount of heave expected (depending on the surcharge pressure) is based on one-dimensional laboratory swelling curves. Since time, site, and budget limitations frequently do not allow complete laboratory testing, empirical correlations are commonly used instead of the one-dimensional laboratory curves. In local studies conducted in 1969 and 1985, these correlations yielded the required general swell model, containing the following independent variables: (a) liquid limit, (b) ratio between the in-situ moisture content and the plasticity limit, (c) in-situ dry density, and (d) vertical surcharge pressure. The first three independent variables constitute clay characteristics. This general swelling model has recently been updated by applying the Excel-solver analysis to new local test results from 352 undisturbed specimens. In addition to updating the model, the paper describes an alternative analysis carried out on the local test results. This alternative analysis is based on the following two-stage operation: (a) conducting a multiple linear regression on the swelling-pressure tests results (i.e., the ASTM 4546 Method C test results) to obtain the swelling pressure correlation with the three clay characteristics mentioned above; (b) performing an additional linear regression on the swelling-percentage test results (i.e., the ASTM 4546 Method B test results) with a single independent variable defined by the given surcharge pressure divided by the predicted value of the swelling pressure, utilizing the correlative equation obtained in the previous stage. Finally a comparison of these two general swelling models indicates a preference for the existing model, generated from the Excel-solver analysis.

Excel-solver, expansive clay, heave, swelling model, swelling percentage, swelling pressure.