Dynamic liquefaction of shear zones in intact loess during simulated earthquake loading

The 2010–2011 Canterbury earthquake sequence in New Zealand exposed loess-mantled slopes in the area to very high levels of seismic excitation (locally measured as >2 g). Few loess slopes showed permanent local downslope deformation, and most of these showed only limited accumulated displacement. A series of innovative dynamic back-pressured shear box tests were undertaken on intact and remoulded loess samples collected from one of the recently active slopes replicating field conditions under different simplified horizontal seismic excitations. During each test, the strength reduction and excess pore water pressures generated were measured as the sample failed. Test results suggest that although dynamic liquefaction could have occurred, a key factor was likely to have been that the loess was largely unsaturated at the times of the large earthquake events. The failure of intact loess samples in the tests was complex and variable due to the highly variable geotechnical characteristics of the material. Some loess samples failed rapidly as a result of dynamic liquefaction as seismic excitation generated an increase in pore water pressure, triggering rapid loss of strength and, thus, of shear resistance. Following initial failure, pore pressure dissipated with continued seismic excitation and the sample consolidated, resulting in partial shear strength recovery. Once excess pore water pressures had dissipated, deformation continued in a critical effective stress state with no further change in volume. Remoulded and weaker samples, however, did not liquefy and instead immediately reduced in volume with an accompanying slower and more sustained increase in pore pressure as the sample consolidated. Thereafter, excess pressures dissipated and deformation continued at a critical state. The complex behaviour explained why, despite exceptionally strong ground shaking, there was only limited displacement and lack of run-out: dynamic liquefaction was unlikely to occur in the freely draining slopes. Dynamic liquefaction, however, remained a plausible mechanism to explain loess failure in some of the low-angle toe slopes, where a permanent water table was present in the loess.