Abstract:
The shrink swell test (Iss) is a commonly used method for estimating the reactivity of soils. The test came into existence in the 1980’s and now underpins a codified approach to the foundation design of lightweight buildings on expansive (reactive) soils in both New Zealand and Australia, in accordance with AS 2870:2011. Indeed, it has been generally accepted by engineering practitioners that the shrink swell test is a simple and economic means of assessing soil expansiveness, which is achieved via several simplifying assumptions that circumvent the measurement of soil suction. In particular, the Iss test estimates the strain change per unit of soil, and assumes that the strain is independent of the initial moisture content. However, recent studies have indicated that the shrink swell index values are not independent of, but are in fact a part-function of, the initial moisture content of soil. Moreover, in many cases, the Iss does not correlate with typical soil index testing values, and analysis of existing datasets demonstrates that the shrink swell test is impacted significantly by a shrink strain bias, and not a swell strain bias, as envisaged by the Iss formula. In this study, twenty-five (25) selected soil samples from three (3) locations within the Auckland region were subjected to a testing program. X-Ray Diffraction, X-Ray Fluorescence and Scanning Electron Microscopy tests were utilised to understand the mineralogical composition of soil samples. The shrink swell index, Atterberg limit and particle size distribution tests were conducted to assess the physical properties of soils. The results of Atterberg limits were plotted against shrink swell index in the form of bivariate scatterplots, and the best-fitting trendlines that were trialled included linear, power-law, logarithmic, polynomial and exponential regressions. The approach taken here was that the trendline with an R² value closest to one (1), was selected. The observed relationships were then graphically compared with existing published relationships and datasets for similarities. The analysis of Auckland datasets and their subsequent comparisons identified significant variation in shrink swell index values with changes in initial moisture content. Further examination of these datasets indicate that shrink swell test has a significant shrink strain bias. The results of this study provide compelling evidence that the shrink swell test and shrink swell index should not be the only guiding factors for foundation design on expansive soils in New Zealand.