Numerical modelling of the behaviour of stone and composite stone columns in soft soils

Abstract

The use of stone columns as a means of ground improvement has been in use for over 40 years in the United Kingdom and Europe. Their primary purpose is to reduce settlement, reduce consolidation time and increase the bearing capacity of soils. Currently the technique is applied to a variety of soil types, cohesive and granular. Soft cohesive soils have shown a tendency towards higher settlements due to the inability of the soil to restrain the lateral movement or bulging of stone columns. Current analytical design methods are based upon the unit concept which considers a stone column to be part of an infinite array of columns. Such methods have proved useful when designing large arrays such as those utilised beneath embankments or large rafts. Columns within the group are restrained equally on all sides and held in the same vertical stress conditions. However, at the edge of large (wide) load areas and in smaller foundation configurations columns are not generally restrained on all sides by other columns and must rely on the soil to provide restraint in the outward facing directions. The behaviour of small foundation configurations is more complex due to this lack of restraint with columns subject to deformation at lower stress levels than those in infinite arrays. This dissertation is concerned with the behaviour of stone columns and proposed composite stone columns installed in soft clay. This research compares the behaviour of small foundations supported by stone columns to behaviour within an infinite array of columns. Specifically the settlement and deformation behaviour of stone columns are considered to identify the main deformation mechanisms and to examine the effect of key design parameters and soft cohesive soils on column performance. A new form of composite stone column was then examined numerically to assess the potential for enhanced column behaviour and settlement reduction. PLAXIS 3D Foundation is utilised with column behaviour represented by the Mohr-Coulomb Perfect Plasticity model and the Hardening Soil model adopted to model soil behaviour. The soft soil profile adopted in this research is the well characterised Bothkennar soft clay site which was formerly the UK geotechnical test bed. The influence of key stone column design parameters, area ratio, column length, column confinement and arrangement, column stiffness, column strength, installation effects and the effect of stiff crust thickness was examined for a combination of foundation types with 432 numerical sensitivity studies conducted. The results reveal that area ratio and column length have a significant impact on the settlement performance of stone columns. Increasing the area ratio was found to reduce the restraint provided by neighbouring columns leading to increased settlement. Increasing column length was found to reduce settlement. When columns were modelled with low area ratios increasing column length had a greater effect on settlement reduction than at higher ratios. The design parameters of area ratio and column length are established as the controlling parameters for the mode of deformation. The mode of deformation was examined utilising settlement inferred deformation ratios (compression and punching) with comparison to total shear strain plots and stress states in the column. Two primary modes of deformation, bulging and punching (including sub-type termed 'block failure') were inferred. Punching failure was inferred for short columns by high punching ratios and low compression ratios with a concentration of shear strain observed at the base of the floating columns. A sub-type of punching, block failure, was inferred from low compression and low punching ratios for closely spaced columns with low area ratios in which the columns act as one unit punching into the underlying soil. Bulging failure was inferred by low punching ratios and high compression ratios coupled with a concentration of shear strain in upper region of the columns. The magnitude of bulging was found to be at its most severe for high area ratios. Bulging as a mode of failure occurred for column length to diameter ratios greater than 4 and area ratios greater than 8. Bulging was found to occur at the weakest of the soil profile which coincides with the top of the lower Carse clay. Consideration was given to a method of reducing the potential for lateral column deformation or bulging by the use of a novel composite column. The deformational characteristics of a stone column were identified for a composite of granular and the experimental Protomix materials. Laboratory testing was carried out to gain an understanding of the cohesive, stiffness and unconfined compressive strength properties of the composite before simulation studies were performed on key design parameters such as area ratio, column length, column confinement and arrangement for a combination of foundation types with 108 numerical analysis sensitivities conducted. The inclusion of a cohesive 'binder' material in the bulging zone was found to reduce settlement for all foundation configurations. Similarly to stone columns area ratio and column length were found to be the design parameters which influenced the results most. The composite stone columns (CSC) offered higher settlement reduction than traditional stone columns (SC). It was discovered that CSC with an area ratio of 8 were able to achieve the same settlement improvement factor as those with a ratio of 3.5 which suggests the columns could offer the same settlement control but with large column spacing's making their use more economical. The settlement inferred deformation ratios (compression and punching) were studied while monitoring the total shear strain field cross sections to examine if composite stone columns would behave similarly to a stone column. It was noted that the same modes of deformation of punching (including block failure) and bulging failure were observed. The increased stiffness in the bulging zone saw the transfer of bulging type effects to a depth below the composite treated zone. It was only observed for high area ratios. The improved settlement behaviour of CSC compared to SC is due to the treatment of the bulging zone by CSC and improved column restraint at depth provided by the soil. Punching failure was found to have a higher magnitude and occur to a deeper depth of 3.6 m compared to SC depth of 2.4 m due to the addition of the composite material. The modes of deformation observed for SC were also observed for the new novel CSC columns. This suggests that the same type of foundations can be used and so avoid the need for reinforcement of the foundations as used with piled foundations.