The principal objectives of this research are to investigate the triggering and failure mechanisms of static liquefaction in loose fill slopes and to study role of soil nails during slope liquefaction. This research consists of three major components, namely centrifuge modeling of static liquefaction of loose fill slopes using Leighton Buzzard (LB) sand, investigation of role of soil nails during liquefaction of loose sand fill slopes in dynamic centrifuge tests and evaluation of liquefaction potential of loose fill slopes using completely decomposed granite (CDG) in both static and dynamic centrifuge tests. In addition, preliminary numerical analyses are carried out to assist in the explanation of results obtained from the centrifuge model tests. Model slopes were formed by compacting...[ Read more ]

The principal objectives of this research are to investigate the triggering and failure mechanisms of static liquefaction in loose fill slopes and to study role of soil nails during slope liquefaction. This research consists of three major components, namely centrifuge modeling of static liquefaction of loose fill slopes using Leighton Buzzard (LB) sand, investigation of role of soil nails during liquefaction of loose sand fill slopes in dynamic centrifuge tests and evaluation of liquefaction potential of loose fill slopes using completely decomposed granite (CDG) in both static and dynamic centrifuge tests. In addition, preliminary numerical analyses are carried out to assist in the explanation of results obtained from the centrifuge model tests. Model slopes were formed by compacting LB sand with water content of 5% and CDG with water content of 11% to an initial relative compaction ranging from 65% to 70%. In static centrifuge tests, ground water was raised or rainfall infiltration was simulated in initially unsaturated slopes. In dynamic centrifuge tests, sand slopes were shaken after slopes had been subjected to limited shallow seepage failure and was submerged under water and CDG slopes were saturated with viscous fluid and shaken at high g.

A sudden and brittle liquefied rapid flow failure was observed and recorded in a 24° loose sand fill slope with a prototype height of 12.9m. This type of failure was only induced by raising the ground water table but was not triggered by rainfall infiltration. During the rapid liquefied flowslide, significant excess positive pore water pressures were measured. It was illustrated that the rapid flowslide was caused by static liquefaction of the loose sand resulting from the strain-softening of the sand under "undrained" conditions. However, the strain-softening behaviour of loose materials was only a necessary but not sufficient condition to induce static liquefaction failure. A trigger was needed. Based on the numerical analysis, the observed fluidised rapid flowslide was triggered by the sliding of the edge of the crest, which had been induced by a seepage flow failure under 'drained' condition.

During the rise of ground water, soil nails helped to reduce wetting-induced volume change in loose sand fill slopes. Shallow seepage failure was not prevented by soil nails. The 15° nailed slope with height of 5.9m was subjected to a shaking with a peak acceleration of 0.16g, which was 8.8 times that applied to the unreinforced slope. A relatively shallower failure was observed in the nailed slope with the measured maximum excess pore pressure reduced by 17.5%, the settlement and the lateral movement of sand mass reduced by 62% and 41%, respectively. Soil nails were illustrated and verified to reduce volume contraction of the slope and to resist soil deformation before slope liquefaction. However, with shaking as a trigger, liquefied flowslide cannot be prevented by soil nails.

Excessive settlements were measured at the 29°, 16m high loose CDG fill slopes during rise of ground water. Surface erosion was induced by heavy rainfall. No flowslide and liquefied flowslide was triggered under both static and dynamic conditions since CDG had a much lower liquefaction potential as compared with that of LB sand. This is consistent with the findings from a 33°, 4.5m high loose CDG fill slope subjected to rise of ground water table and rainfall infiltration in the field. It is very unlikely, if not possible to induce liquefied flowslide in loose CDG fill slope but shallow non-liquefied flowslide is possible.