The problem of developing efficient and economical liquefaction remediation methods for existing buildings has not been studied extensively. This thesis describes dynamic centrifuge tests undertaken to try and increase understanding of three such remediation methods: containment walls, cemented zones and inclined, non-loadbearing micro-piles. Centrifuge tests with a simple SDOF frame structure founded on both unimproved and improved soil profiles were carried out for direct comparison.
One of the main outcomes of this research was to show that for a remediation scheme to be successful in reducing structural settlements it is necessary that the remediated zone extends through the full depth of the liquefiable layer. Alternatively, it may be sufficient to ensure that the depth of remediation is greater than the expected depth of free field liquefaction.
Both stiff and flexible (geomembrane) containment walls were found to be very effective in reducing both co-seismic and post-seismic structural settlements, if drainage from the surface of the contained soil was prevented. The containment walls restrained lateral movement of the soil below the structure and also limited volume change in this contained soil. In the case of the flexible walls, lateral soil movement was restrained by hoop stress in the membrane. Softening of the contained soil was not prevented, so no increased response of the structure was caused by the presence of the containment walls, assuming the connection detail between the walls and the structure was designed correctly. A cemented zone beneath the structure was found to be very efficient in reducing structural settlements but may have caused slightly increased structural responses. More higher frequency accelerations were transmitted to the structure than in the unimproved case. No conclusive evidence for the micro-piles having a significant beneficial effect was obtained
Predictions of structural response from simple SDOF and 2DOF analytical models for structures founded on dry, saturated and remediated soil were compared to centrifuge data to assess the effectiveness of such methods. It was found that these simple models can predict structural response reasonably well and can capture the effects of soil softening. The importance of use of the correct soil parameters and input accelerations was highlighted, which suggests that use of averaged design spectra, not customised to the individual site conditions will not predict structural response realistically.
This research offers valuable information about the most important aspects of behaviour of the remediation methods studied. The results obtained can help in the design of liquefaction remediation schemes for existing buildings.