BEHAVIOUR OF RANDOMLY DISTRIBUTED FBER-REINFORCED SAND IN SHALLOW FOUNDATIONS

Abstract

Ground improvement in weak soils has become necessary in view of heavy loads imposed by industrial structures, storage tanks and high-rise buildings. Due to scarcity of good land, one has to build on marginal soils or on filled up soil. Reinforced soil is one of the most popular and fastest growing techniques for improvement of poor soils. Much work has been done on bearing capacity improvement for shallow foundations using planar reinforcement. However, limited study has been carried out on use of randomly distributed fiber-reinforced sand (RDFS) in shallow foundations. Fiber reinforcement may have considerable effect on the bearing capacity improvement same as that by using oriented geogrid layers. RDFS is an emerging technology and it has been successfully used in variety of applications such as slope stabilization, road sub-grade and sub-base. The available literature indicates that this is relatively a simple technique for ground improvement which may have enormous potential for economical solutions to many geotechnical problems. But fiber-reinforced soil still has limited case histories which suggest the additional research to be done in this area. The objective of present research programme is to explore the application of discrete fibers for bearing capacity improvement and reduction in settlement, tilt & horizontal displacement of footings. In this research work comprehensive experimental study has been carried out on fiber reinforced sand, which includes the study on strengthdeformation characteristics of RDFS using triaxial set up and model footing tests under central, vertical, eccentric and inclined loads. To study the strength - deformation characteristics of RDFS, drained triaxial tests were conducted on polypropylene fibers of different deniers (1denier = mass in grams per 9000 mlength of fiber = 1.11 x 10"7 kg/m) and lengths. Inthese tests it was observed that with addition of fibers in sand, there is increased peak shear strength, improvement in the post-peak response. In most of cases strain hardening effect at large strains (i.e. more ductile behaviour) was observed which suggests its suitability in foundation material where large deformations are to be tolerated such as earthquake resistant geotechnicalstructures and footings subjected to eccentric-inclined loads. In RDFS, the failure does li not take place even at strain of20% or more. Using experimental data, hyperbolic stressstrain parameters were determined and it was found that Kondner's hyperbolic stressstrain relationships are valid for RDFS. Model footing tests under central and vertical load were conducted in test box of size 800 mm long, 77mm wide and 400 mm deep. The Solani river sand reinforced with different type ofpolypropylene fibers (monofilament and fibrillated) and polypropylene mesh elements (cut from Netlon geogrid CE 121 and Netlon advanced turf system: NAF) was used in the study. Amodel footing ofsize 75 mm x 75 mm was used. To study the effect of relative density, tests were conducted at relative density of 30%, 50% and 70% for sand and RDFS. Whole tank was filled with RDFS. No lateral movement of side walls of test box was allowed by providing stiffeners all around it. Thus plane strain conditions prevailed and footing acted as strip footing of 75 mm width (B). Significant improvement in the bearing capacity and reduction in the settlement was observed due to the inclusion ofdiscrete fibers in the sand beneath the footing. It was found that bearing capacity increased and settlement reduced with the increase in the fibre content. Improvement in pressure - settlement behaviour and bearing capacity of RDFS vary significantly for different discrete fibers and mesh elements. Lowest improvement is given by mesh elements from Netlon CE 121 and highest improvement is given by fibrillated fibers. Length or aspect ratio (tj = length/ equivalent diameter or lateral dimension) offiber or mesh may have an effect on bearing capacity but it is primarily the type offiber or mesh element which matters most. Length appears to be more important factor than aspect ratio. The 6 denier 10 mm long (tj = 325) monofilament fibers give much less improvement compared to 20 denier 20 mm long (rj = 350) monofilament fibers having similar aspect ratio. Pressure - settlement curves of 6 denier 20 mm, 20 denier 20 mm, 20 denier 50 mm are almost similar. "NAF" mesh elements also gave relatively good improvement comparable to thin monofilament fibers but not better than fibrillated fibers. Thus "mesh elements perform better than fiber" is not necessarily true as reported by many investigators. However, mesh elements perform better than straight fiber cut from same mesh is true. Fibrillated fibers are though straight but during mixing these forms mesh like structure. Fibrillated fibers have shown very good improvement even at very low values of settlement ratio (settlement/width of footing) and at in high settlements, a strain hardening behaviour was observed. At a fiber content of1% by weight, bearing capacity ratio i.e. BCR (BCR is defined as ratio ofbearing capacity of RDFS to bearing capacity of sand alone at 10% settlement ratio) was around 10 at relative density of30% for 1000 denier 50 mm long fibrillated fibers. In general it can be concluded that fibrillated fibers perform best amongst all fibers/ mesh elements used in present study. Model footing tests were conducted at different relative densities on RDFS to study effect of density on improvement. Consistent improvement is shown at all relative densities. However BCR decreases with increase in relative density but absolute increase inbearing pressure is much high at high relative density. To find optimum zone of reinforcement below footing, tests were conducted on varying width and depth of RDFS zone. Providing RDFS below footing in zone of 2B depth and3B width found most effective. It would be more beneficial to reinforce sand in shallow depths with high fiber content as compared to low fiber content for deeper depth. Even providing RDFS zone of 0.5B depth and 2B width below footing is also very effective and BCR was found to increase upto 2.8 for 1% fiber content at 30% relative density. A comparison of RDFS with horizontally placed geogrid reinforcement was also studied. Optimum and maximum increase in bearing capacity using only geogrid reinforcements was found by using 3 layers of geogrid of sizes 75mm x 375 mm each placed at spacing of 0.33B below footing. In this case bearing capacity ratio increased to 3.2 which is less than 3.3 obtained by using0.5% of fibrillated fibers in 2B width and IB depth only. Cost of 3 geogrid layers is about six times compared to 0.5% of fibrillated fibers in 2B width and IB depth. Cost of single layer planar reinforcement (BCR - 2) is eight times compared to 0.25% fibers used in 2B width and 0.5B depth (BCR=2.1). Such a high increase in bearing capacity of cohesionless soil can safely permit use ofshallow foundations on RDFS in place ofdeep foundation or other such costly options and prove to be aneconomical solution for foundations subjected to heavy loads imposed by industrial structure, high rise buildings, storage tanks etc. IV An analytical method based on constitutive laws of RDFS (Kondner's hyperbola) has been proposed to predict pressure-settlement behaviour of strip footing resting on fiber reinforced soil. This analytical method is able to predict pressure-settlement curve upto 2/3 of ultimate load. Predicted pressure settlement curve using this method compare very well with reported data in literature. Tests conducted under submerged conditions record a decrease in bearing pressure of about 40% nearly at all settlement ratio which was expected due to decrease in unit weight of sand due to submergence. Under submerged condition also RDFS is equally effective as in dry condition (thus submergence under water has no adverse effect on improvements gained by RDFS). Model tests have also been conducted on model strip footing (75 mm x 75 mm) resting on unreinforced and reinforced Solani river sand under eccentric-inclined load in two dimensional tank (plane strain condition). Tests were conducted at eccentricity ratio e/B = 0.0, 0.1, 0.2, load inclinations i = 0°, 10°,20° and relative density of 30%, 50%and 70%. Fibrillated fibers of 1000 deniers 50 mm long have been used in all tests on RDFS. Some tests were conducted on 1000 deniers 20 mm and 360 deniers 20 mm fibrillated fibers also. The pressure-settlement, pressure-horizontal displacement and the pressuretilt curves had been obtained for each model test. Pressure-settlement behaviour of footings resting on unreinforced sand under eccentric-inclined load was very poor and failure took place at very small deformations. In case of RDFS, pressure-settlement behaviour improved substantially. There was no sudden decrease in bearing pressure with increase in settlement. BCR keeps on increasing as eccentricity and inclination of load increases. Tilt and horizontal displacement decreased substantially. It suggests that for eccentric-inclined loads, beneficial effects of RDFS increased further in comparison to central-vertical loads. The depth and width of the fiber-reinforced zone had been varied and it was found that most optimum zone for eccentric-inclined load is 6 B wide and 1 B deep below footing. The findings of model test results have been utilised to develop a regression model for strip footing on RDFS in non-dimensional form. Comparison of predicted values of bearing capacity ratio, settlement and tilt for footings on RDFS, has shown good agreement with their respective values of model test. To validate models developed and know the influence of size of footing on improved pressure - settlement - tilt behaviour due to RDFS, a series of tests were conducted on 50 mm, 100 mm and 150 mm wide footing in plane strain condition. Results showed that proposed regression model in non-dimensional form predicts very well the BCR and settlement for 50 mm, 100 mm and 150 mm wide footing. For eccentrically and obliquely loaded footings also and results showed that proposed model predicts very well the BCR, vertical settlement and tilt. These findings suggest that nonlinear multiple regression models developed in present research work in the nondimensional form based on model tests can be used for design of prototype foundations resting on randomly distributed fiber-reinforced sand.