An experimental study on geosynthetic reinforced soil walls under seismic loading

Abstract

Reinforcement of soil enables a soil slope or wall to be retained at steeper angles than the soil on its own could manage. Geosynthetic Reinforced Soil (GRS) wall structures incorporate geosynthetic reinforcement (a strong and flexible engineering plastic grid) layered horizontally into soil walls to stabilise and/or retain the soil. They take less time to construct at a reduced cost (with savings of up to 50%) and perform better in earthquakes than their conventional retaining wall counterparts such as large concrete based gravity and cantilever type retaining walls. Further, GRS walls tend to be more environmentally sustainable, with reduced carbon emissions and embodied energy.

In countries like Japan which has a high earthquake risk, GRS technology is used for important and vulnerable lifeline assets such as high-speed railways. New Zealand has a similar earthquake risk, yet the use of GRS technology has been limited. To help increase the uptake of GRS technology on New Zealand's roads and railways, a better understanding of GRS behaviour in the New Zealand earthquake environment is required.

To investigate GRS behaviour under earthquake loading, a series of seven reduced-scale GRS model walls were subjected to earthquake shaking on the University of Canterbury shake-table. The models were 0.9 m high, 2.4 m long, and 0.8 m wide, and comprised dry dense sand, reinforced by five layers of stiff geosynthetic reinforcement and rigid front face. Scaling laws applied to seismic shaking were employed so that the wall effectively represented a prototype 4.5 m high wall. The length of reinforcement, and the slope of the wall face was varied to investigate these parameters' influence on wall stability. Each model was shaken with increasing intensity until failure occurred. During testing, acceleration and displacement measurements, as well as three high-speed cameras were used to plot the progression of how the model deformed. This deformation generally occurred with the wall collapsing over its base and sliding forward. The results of the study demonstrated that GRS walls built with longer geosynthetic reinforcement and shallower slopes were more stable and able to withstand stronger earthquakes. 

Finally, advanced image analysis software was used to show deformation previously undetectable by eye at relatively low amplitudes of shaking. These results highlight the relative ductility of GRS walls, and have implications to the way GRS systems are designed.

Technical Abstract

Reinforcement of soil enables a soil slope or wall to be retained at angles steeper than the soil material's angle of repose. Geosynthetic Reinforced Soil (GRS) systems enable shortened construction time, lower cost increased seismic performance and potentially improve aesthetic benefits over their conventional retaining wall counterparts such as gravity and cantilever type retaining walls. Experience in previous earthquakes such as Northridge (1994), Kobe (1995), and Ji-Ji (1999) indicate good performance of reinforced soil retaining walls under high seismic loads. However, this good performance is not necessarily due to advanced understanding of their behaviour, rather this highlights the inherent stability of reinforced soil against high seismic loads and conservatism in statis design practices.

This is an experimental study on a series of seven reduced-scale GRS model walls with FHR facing under seismic excitation conducted using a shake-table. The models were 900 mm high, reinforced by five layers of stiff Microgrid reinforcement, and were founded on a rigid foundation. The soil deposit backfill was constructed of dry dense Albany sand, compacted by vibration (average Dr=90%). The influence of the L/H ratio and wall inclination on seismic performance was investigated by varying these important design parameters throughout the testing programme. The L/H ratio ranged from 0.6-0.9, and the walls were primarily vertical except for one test inclined at 70º to the horizontal.

During testing, facing displacements and accelerations within the backfill were recorded at varying levels of shaking intensity. Mechanisms of deformation, in particular, were of interest in this study. Global and local deformations within the backfill were investigated using two methods. The first utilised coloured horizontal and vertical sand markers placed within the backfill. The second utilised high-speed camera imaging for subsequent analysis using Geotechnical Particle Image Velocimetry (GeoPIV) software. GeoPIV enabled shear strains to be identified within the soil at far smaller strain levels than that rendered visible by eye using the coloured sand markers. The complementary methods allowed the complete spatial and temporal development of deformation within the backfill to be visualised.

Failure was predominantly by overturning, with some small sliding component. All models displayed a characteristic bi-linear displacement-acceleration curve, with the existence of a critical acceleration, below which deformations were minor, and above which ultimate failure occurs. During failure, the rate of sliding increased significantly.

An increase in the L/H ratio from 0.6 to 0.9 caused the displacement-acceleration curve to be shallower, and hence the wall to deform less at low levels of acceleration. Accelerations at failure also increased, from 0.5g to 0.7g, respectively. A similar trend of increased seismic performance was observed for the wall inclined at 70º to the horizontal, when compared to the other vertical walls.

Overturning was accompanied by the progressive development of multiple inclined shear surfaces from the wall crest to the back of the reinforced soil block. Failure of the models occurred when an inclined failure surface developed from the lowest layer of reinforcement to the wall crest. Deformations largely confirmed the two-wedge failure mechanism proposed by Horii et al (2004).

For all tests, the reinforced soil block was observed to demonstrate non-rigid behaviour, with simple shearing along horizontal planes as well as strain localisations at the reinforcement or within the back of the reinforced soil block. This observation is contrary to design, which assumes the reinforced soil block to behave rigidly.