The significance of boundary conditions in dynamic centrifuge modelling

Berrak Teymur, Cambridge University
Geotechnical Engineering Group


There have been many devastating earthquakes throughout history. Research on earthquake effects on different soil-structure systems has increased over the years following these destructive earthquakes. Dynamic centrifuge modelling is a useful tool for studying earthquake engineering problems.

In a centrifuge, the space available to model real situations is not infinite, and it is necessary to enclose the model within the finite boundaries of a container. The boundary effects of the soil container are important and can lead to inaccurate simulation of a field situation that has infinite lateral extent. The Equivalent Shear Beam (ESB) container was built from dural and rubber layers to achieve the same dynamic response as the soil sample, as this is one method of minimising boundary effects.

A series of centrifuge tests involving loose and dense, dry and saturated models of homogeneous horizontal sand layers has been carried out to investigate the effects of the end-walls of the ESB model container on soil behaviour. In both saturated and dry sand models it was seen that amplification of acceleration occurs at the base of the model close to the end-wall. Towards the surface of the soil in saturated models, acceleration attenuates whereas in dry sand it amplifies. In saturated models as excess pore-pressures build up, the response of the centre of the model gets progressively more different from that of the end-wall. This is to be expected as the box stiffness is matched to that of the non-liquefied soil column. It was seen that when the relative density was around 50%, close to that of the design soil layer, the change in behaviour towards the end-wall was minimal.

Experiments carried out with an in-flight miniature CPT apparatus, show that there is a difference in the soil stiffness before and after the earthquake loading due to the presence of rough complementary shear sheets placed at the end walls. The rough sheets provide shear stresses in the vertical direction in the soil following earthquake loading that oppose the settlement undergone by the rest of the soil sample. Thus contradictory requirements arise that the boundary walls should be smooth under static loading and rough during the dynamic loading. The data acquired from the miniature CPTs was used to assess liquefaction potential of the model based on empirical charts. This comparison led to the conclusion that these charts based on field data should be revised with the available centrifuge data to include various liquefaction ratios at various depths.

Harmonic wavelet analysis was applied to accelerations measured during the Kocaeli (Turkey) earthquake. The presence of two peaks with different frequency contents in the earthquake signals was observed. The comparison of wavelet maps of acceleration at two different stations indicate that similar structures in different regions may respond differently during the same earthquake. This emphasises the importance of determining the site specific response.

It was concluded based on the harmonic wavelet analysis of the acceleration signals that the boundary of the ESB functions best with dense dry sand models whose response changes little with time. Saturated models in which liquefaction affects the soil stiffness show significant non-uniformity of behaviour. The stiffer soil at the base of the box matches the dynamic response of the container better than that at the surface.