Characterization of expansive clay behavior using centrifuge and field evaluations




Armstrong, Christian Philip

Journal Title

Journal ISSN

Volume Title



Subsequent to the development of a centrifuge-based test methodology at the University of Texas at Austin, an extensive research program was conducted to quantify the parameters that govern the swelling behavior of expansive soils and to correlate field measurements of heave and moisture fluctuations with predictions from laboratory testing. The overall research was grouped into five self-contained research components. The first research component examined the effect of the initial conditions in centrifuge soil samples on the magnitude and time-history of swelling. The centrifugation of soil specimens was found to not affect the magnitude of swelling, further validating the centrifuge technique as practical and accurate. The second research component examined the effect of soil fabric on the swelling of laboratory-reconstituted and field-sampled clay specimens. Soil fabric was found to affect the time-history of swelling for laboratory-reconstituted specimens and to affect both the time-history and magnitude of swelling for field-sampled specimens. These results can be attributed to the field-sampled specimens containing a higher amount of peds and micro-void features, as revealed using scanning electron microscope micrographs. These micro-void features are removed during soil homogenization in the laboratory-reconstituted samples. The third research component examined the heterogeneity on a regional and local scale of two clay soils commonly found in Central Texas. The results indicated that soil samples derived from geologically older parent material tended to swell more than soils derived from geologically younger parent material and that the method of soil sampling affected the magnitude of swelling. The fourth research component examined the field behavior at a site involving an expansive soil subgrade and was instrumented using moisture content and suction sensors. Moisture fluctuations were found to be significant regardless of the pavement surface, and soil-water retention curves were found not to vary significantly with depth. The heave measured at the site was less than that predicted by centrifuge test results due to volumetric strains and desiccation cracks. The final research component examined the large-scale swelling behavior of a column test, prepared to represent a field site. Results from the large column test were compared to the heave predicted using centrifuge test results and the previous field site. The large column was found to behave similarly to the predictions from centrifuge test results, but the infiltration occurred over a longer time frame than the field site. The difference in infiltration rates between the large column and field site was attributed to desiccation cracking at the field site. Overall, the five research components involving expansive soil behavior indicate that the field monitored swelling can be predicted using laboratory centrifuge tests when properly accounting for initial conditions, soil fabric, and spatial variability.


LCSH Subject Headings