Small-scale modelling of plant root systems using 3D printing, with applications to investigate the role of vegetation on earthquake-induced landslides

Abstract

Vegetation has been previously proposed as a method for protecting artificial and natural slopes against shallow landslides (e.g. as may be triggered by an earthquake); however, previous research has concentrated on individual root soil interaction during shear deformation rather than the global slope behaviour due to the extreme expense and difficulty involved in conducting full-scale field tests. Geotechnical centrifuge modelling offers an opportunity to investigate in detail the engineering performance of vegetated slopes, but its application has been restricted due to the lack of availability of suitable root analogues that can repeatably replicate appropriate mechanical properties (stiffness and strength) and realistic 3D geometry. This study employed 3D printing to develop a representative and repeatable 1:10 scale model of a tree root cluster (representing roots up to 1.5 m deep at prototype scale) that can be used within a geotechnical centrifuge to investigate the response of a vegetated slope subject to earthquake ground motion. The printed acrylonitrile butadiene styrene (ABS) plastic root model was identified to be highly representative of the geometry and mechanical behaviour (stiffness and strength) of real woody root systems. A programme of large direct shear tests was also performed to evaluate the additional strength provided by the root analogues within soil that is slipping and investigate the influence of various characteristics (including root area ratio (RAR), soil confining effective stress and root morphology) on this reinforcing effect. Our results show that root reinforcement is not only a function of root mechanical properties but also depends on factors including surrounding effective confining stress (resulting in depth dependency even for the same RAR), depth of the slip plane and root morphology. When subject to shear loading in soil, the tap root appeared to structurally transfer load within the root system, including to smaller and deeper roots which subsequently broke or were pulled out. Finally, the root analogues were added to model slopes subjected to earthquake ground motion in the centrifuge, where it was revealed that vegetation can substantially reduce earthquake-induced slope deformation in the soil conditions tested (76% reduction on crest permanent settlement during slippage). Both the realistic 3D geometry and highly simplified root morphologies, as characterised mechanically by the shear tests, were tested in the centrifuge which, despite exhibiting very different levels of additional strength in the shear tests, resulted in very similar responses of the slopes. This suggests that once a certain minimum level of reinforcement has been reached which will alter the deformation mechanism within the slope, further increases of root contribution (e.g. due to differences in root morphology) do not have a large further effect on improving slope stability.