Cyclic response of fibre reinforced dense sand

Abstract

Listen

Translate article

Fibre reinforcement has proven to be an adequate technical solution for soil improvement where its applications include slopes, railways, highways and, more recently, wind turbines foundations or any structure subjected to the effect of cyclic loads. A few studies have investigated the cyclic behaviour of fibre reinforced soils, and they mostly focus on loose materials and liquefaction. In this sense, this study investigated the cyclic behaviour of a fibre reinforced dense sand under undrained triaxial conditions. Drained and undrained monotonic and undrained cyclic triaxial tests were performed with varying initial effective stresses (20 to 200 kPa) and cyclic stress ratios (0.1 to 0.5). The soil, a uniform fine sand, was reinforced with 50 mm long polypropylene fibres and compacted to a dense state (Dr = 90%). Fibres contribution, as expected, improved sand monotonic behaviour under compression with material’s strength parameters shifting from ϕ = 39.8° and c’ = 0 kPa to ϕ = 53.5° and c’ = 0 kPa, under drained conditions, and from ϕ = 37.2° and c’ = 0 kPa to ϕ = 43.9° and c’ = 41.6 kPa, under undrained conditions. Accordingly, under cyclic loading conditions fibres also enhanced mixtures strength which is depicted by the increase of cycles until failure and a meaningful reduction of both cyclic and permanent pore water pressures development. Also, in spite of reinforcement inclusion or not, dense specimens experienced the cyclic mobility phenomena at failure, namely, samples were able to withstand loading cycles even after great mean effective stress reduction. Fibres inclusion turned dense sand specimens stiffer than the unreinforced ones when subjected to cyclic loads. Finally, the application of a constitutive model was investigated to predict the fibre-reinforced dense material behaviour. The applied model was not capable of predicting the behaviour of the investigated fibre-reinforced dense sand under the whole range of studied variables. The response for higher CSR values was better captured by the model.