A Critical Analysis of Methods for Testing Reactive (Expansive) Soils on Waiheke Island

Abstract

The methods outlined in the current New Zealand Standards (NZS 3604) for classifying reactive (expansive) soil sites is disputed. In recent years, the methods used to accurately predict characteristic surface movement have been in question. This research focusses on residually weathered soils from the Waipapa (Waiheke) Group and additional datasets from previous studies discussed and used for comparison. A sampling and testing program was undertaken on soil samples collected from eight different sites across Waiheke Island. One disturbed sample from each site was collected to test for Atterberg Limits (AL) and linear shrinkage (LS). One undisturbed sample from each site was collected to test for the shrink strain, swell strain and shrink swell index (ISS). Upon completion, the results of the tests were critically analysed. The AL and LS tests appear to provide reliable values, however, unlike the ISS, there is no recognised formula for the AL and LS to be used to predict the characteristic surface movement (ys) and corresponding site class. The AL and LS results were plotted against the ISS and shrinkage index in an attempt to devise a reliable correlation between the soil parameters. The strongest trendline, associated equation and R2 value were added to each graph. For a correlation to be deemed reliable, R2 > 0.80. The equations were used to calculate a predicted ys and these were compared to the ys calculated from the ISS. There were no correlations that exceeded R2 > 0.80 and the predicted ys were lesser than the actual ys values (from the ISS). The results of the shrink swell test were examined to determine its reliability. The parameters were plotted against each other including, the initial moisture content (IMC) against the ISS, the ISS against each individual test strain, and each individual test strain against the IMC. The plots indicated that the ISS is almost entirely governed by the shrink strain, which has an initial moisture bias. The shrink swell test should not be relied upon by engineers to accurately predict the ys and site soil class. This research indicated that further work should be undertaken to change the way reactive soil sites in New Zealand are classified. A theory is proposed for consideration that relates the AL and LS to the ys, factoring in climate and in-situ conditions.