Soil Performance For Large Scale Soil-Pipeline Tests
The strength and dilation properties of dry and partially saturated sand at peak and critical state are measured and compared in terms of the matric suction measured for partially saturated sand. Techniques are developed for the movement, placement, compaction and measurement of large quantities of soil with tight control of water content, w, and dry unit weight, lambda-dry. Tactile pressure sensors are used to measure pipe-soil interaction stresses reliably. Direct shear (DS) tests performed on three different glacio-fluvial sands with DS box dimensions of 60, 100 and 300 mm show that the conventional 60 mm apparatus produces results that are consistently higher in terms of peak strength and dilatency than those of the larger boxes. Based on an evaluation of the test results, a modified 100 mm apparatus, consistent with the design proposed by Lings and Dietz (2004), is recommended for future DS testing. This DS apparatus gives high quality data, and is validated through favorable comparison with flow rules relating strength and dilation at peak to critical state strength. The desorption soil water retention curve (SWRC) measured with Tempe cells is used to predict an adsorption SWRC, which compares favorably with tensiometer measurements. The effects of matric suction on the peak angle of shear resistance and dilatency are quantified by comparing DS test results for dry and partially saturated sand of identical composition and lamba-dry. Apparent cohesion at peak stress is found to be dependent on dry for partially saturated sand, and is related to increased dilatency. It appears that matric suction increases the interference among the sand particles, thus increasing the work against volumetric expansion during shear failure. Methods of soil placement, compaction and measurement for large scale tests are shown. The mean values of w and lambda-dry for tests with up to 320 measurements taken are shown to have 95% confidence intervals of 0.04 - 0.10% and 0.07 - 0.12 kN/m3, respectively. A comparison of nuclear gage and Selig density scoop measurements of lambda-dry shows that the density scoop overestimates lambda-dry for dry sand, and is in favorable agreement with lambda-dry determined with the nuclear gage for partially saturated sand. A comparison of nuclear gage and ASTM D2216 (ASTM, 2003d) methods for measuring w shows that the nuclear gage systematically underestimates w in the soil mass and is less precise than the ASTM method. Tactile pressure sensors are shown to be suitably accurate and versatile for measurement of normal stresses in large-scale laboratory testing. A two-layer system of Teflon(r) sheets is shown to be effective in protecting the sensors from shear stress effects, and methods for sensor calibration and measurement are proposed for minimizing time-rate of loading and creep effects.
dissertation or thesis