Numerical analysis of a geosynthetic-reinforced piled load transfer platform – Validation on centrifuge test

https://doi.org/10.1016/j.geotexmem.2014.07.012Get rights and content

Abstract

Soft soil improvement techniques using a network of rigid inclusions and geosynthetic reinforcement are investigated to improve our understanding of load transfer mechanisms towards piles. The physical modelling of the system consists in simulating fictional soft soil settlement through downward displacement of a perforated tray above a network of rigid piles placed in the centrifuge swinging basket. Tests are used to validate the results of the numerical study.

Elasto-plastic and hypoplastic constitutive models have been used to predict the behaviour of the granular mattress, which simulates a Load Platform Transfer (LPT). A two-dimensional, axisymmetrical model has been adopted, which fulfils the validation on the experimental test and the time needed for calculation.

The results of the parametric studies show that load transfer increases with mattress thickness and closer pile spacing. Geosynthetic deflection is reduced when load transfer is high.

Introduction

A technical solution to reinforce soft soil consists in using a network of rigid inclusions (Fig. 1). The square mesh of the inclusions is characterized by s, the center-to-center pile spacing. These inclusions possibly reach a rigid substratum. A granular mattress, with a thickness, H, and a density, ρd, is laid on the reinforced soft soil. Inside the mattress, the shearing mechanisms between the grains and the arching effects between the piles can transfer part of the load (term A) directly towards the inclusions. The mattress behaves like a Load Platform Transfer (LPT). To enhance horizontal reinforcement further, a geosynthetic fabric is inserted at the base of the granular mattress. When stretching, the geosynthetic transmits an additional load towards the pile (term B). This load transfer is called the membrane effect (Le Hello and Villard, 2009) The remaining load, only, then, applies on the soft soil (term C). The proportion of terms A, B and C depend on mattress thickness, pile spacing, surcharge, compressibility of the soft soil and secant stiffness of the geosynthetic Ja.

Due to such complexity, numerical modelling has been performed to obtain more information on the load distribution within the mattress and carry out parametric studies (Le Hello and Villard, 2009, Borges and Marques, 2011, Han et al., 2012, Jennings and Naughton, 2012, Nunez et al., 2013). Although numerical models tend to underestimate strain within geosynthetic fabrics, the use of numerical tools is very useful. Besides the contribution they bring to the analytical models used for designing reinforcement of piled embankments (i.e., British Standard BS 8006 (1995) and EBGEO (2011) for instance), they can be used to analyse additional features like, for example, the influence of non-uniform loading or deformation and pile moments. The plane strain configuration and, more recently, the 3D calculation are generally used for computing. The present research is based on the assumption that the embankment height above the mattress is constant, i.e., located far from the slope. Jenck et al. (2009a) have shown that, for an inclusion placed far away from the slopes of the embankment, 2D-axisymetric modelling satisfactorily agrees with the 3D case. The purpose of this paper is to validate a numerical model based on the results obtained using centrifuge testing on prototype scale.

Depending on the studies, different constitutive models of the granular LPT have been used. For example, granular LPT behaviour has been modelled using an elastic-perfectly plastic model with a Mohr Coulomb type failure criterion (Han and Gabr, 2002). Jenck et al. (2009b) or Plaut and Filz (2010) have modelled soft soil using a linear elastic constitutive model. Borges and Marques (2011) have used a Cam Clay constitutive model.

Blanc et al., 2013, Blanc et al., 2014, Okyay et al., 2014 and Rault et al. (2010) have described centrifuge experiments using a mobile tray device, which simulates soft soil settlement through the downward movement of a perforated tray above a network of inclusions. The aim of this paper is to present the findings of the numerical investigations carried out to model and validate the test results achieved using the experimental device developed by Okyay et al. (2014) and then by Blanc et al. (2013).

First, centrifuge modelling principles are reminded and the specific experimental device is briefly described. Two and three-dimensional models are compared to optimize the calculation. The choice of the model materials is discussed. The numerical modelling is then described. Two constitutive models: the Hardening Soil (Schanz et al., 1999) and the hypoplastic model (Kolymbas et al., 1995, Gudehus, 1996) have been adopted for the study of the behaviour of the granular mattress. The comparison between experimental and numerical results for the same geometry allows for the validation of the numerical simulation and the determination of the best constitutive model. Finally, some numerical parametric studies are conducted on H, s, q0, the mattress initial void ratio e0 and secant stiffness of the geosynthetic Ja.

Section snippets

Physical modelling

Centrifuge modelling makes it possible to study geotechnical structures using small scale models subjected to identical stress levels. For a 1/N scale model, the centrifugal acceleration is equal to N times the standard earth gravity acceleration g (N = 20 here). The transposition between the prototype (full scale) and the model (small scale) is guaranteed by the scaling factors presented in Table 1 (Phillips, 1869, Garnier et al., 2007). According to centrifuge principle, the scaling factors

Basic concept

The Plaxis piece of software (Brinkgreve and Vermeer, 2003) is used to model centrifuge small scale test results. The finite element method makes it possible to adopt different constitutive models ranging from the basic elasto-perfectly plastic model to more complex ones. The surcharge is applied homogeneously (1) on the top of the mattress (2), above the geosynthetic (3) and both on the tray (4) and the pile (5) (Fig. 3). Vertical and radial axes are defined from the pile centre: z is the

Numerical modelling results

Although arching is overestimated and creep is not taken into account by the numerical model, the comparison shows that the numerical results reasonably agree with the experimental results. A deeper analysis may give more information on the mattress and the geosynthetic. The numerical analysis provides information on load transfer and differential settlement evolution, on the one hand, and on the geosynthetic behaviour, on the other hand. Movements within the mattress are known at any point of

Geometrical configurations

The influence of the thickness H and the pile spacing s are studied and compared with the physical model. The results are plotted using EF and ΔωCP (Fig. 13). Fig. 13(a) shows the results of the parametric study carried out on H (0.7 m, 1.0 m and 1.8 m), with s constant (s = 2.0 m) and Fig. 13(c) the results of the parametric study carried out on s (2.0 m, 2.8 m and 4.0 m), with H constant (H = 1.0 m).

Like for H = 1.0 m (Fig. 7(a)), the numerical values of EF start from α = 4.91% (s = 2.0 m)

Conclusion

The aim of this paper is to present the findings of the detailed investigation carried out to study the geosynthetic reinforcement of the load transfer mattress located above a network of rigid pile using a finite element model. The comparison between 2D models and a 3D model reveals that the 2D-axisymmetrical model is the best available tool to model load transfer inside the mattress. The behaviour of the granular mattress is modelled using a hypoplastic constitutive model, which takes the

References (36)

  • P. Ariyarathne et al.

    Effect of geosynthetic creep on reinforced pile-supported embankment systems

    Geosynth. Int.

    (2013)
  • G. Baudouin et al.

    Caractérisation mécanique d'un mélange de sables d'Hostun

  • M. Blanc et al.

    Geosynthetic reinforcement of a load transfer platform: centrifuge parametric study of piled embankment

    Geosynth. Int.

    (2014)
  • L. Briançon

    Renforcement des sols par inclusions rigides – Etat de l'art

    (2002)
  • R.B.J. Brinkgreve et al.

    Plaxis v8, manuel de référence

    (2003)
  • BS8006

    Code of Practice for Strengthened/reinforced Soils and Other Fills

    (1995)
  • BS8006-1

    Code of Practice for Strengthened/reinforced Soils and Other Fills

    (2010)
  • J.-C. Dupla et al.

    Caractérisation mécanique à l'appareil triaxial du mélange de sable d'Hostun utilisé en centrifugeuse en tant que matelas de transfert de charge (A.S.I.Ri No. 2-08-3-07), Thème 3: Essais de laboratoire et en centrifugeuse

    (2008)
  • Cited by (52)

    • Three-dimensional physical modeling of load transfer in basal reinforced embankments under differential settlement

      2023, Geotextiles and Geomembranes
      Citation Excerpt :

      The maximum strain in the geogrid was concentrated at the edge of the pile cap, intensely concentrated at the lower layer and was more well-proportioned on the main strip that bridged two adjacent caps (Ye et al., 2020). Girout et al. (2014) and Ye et al. (2020) concluded that the maximum tensile strains were significantly greater in the lower layer than in the upper layer. Thus, adding the geogrid sheets induced a minor effect on the pile load transfer efficiency.

    • Geosynthetic reinforced piled embankment modeling using discrete and continuum approaches

      2021, Geotextiles and Geomembranes
      Citation Excerpt :

      In FLAC3D, the geotextile is modelled by an isotropic linear elastic material with E is the Young's modulus and t is the thickness of the geotextile. A Poisson's ratio equal to 0.3, similar to the studies of Han and Gabr (2002), Liu et al. (2007) (Tano et al., 2016; Tano et al., 2017; Girout et al., 2014), and Pham et al. (2018) is considered. For comparison, two types of elements (‘liner’ and ‘geogrid’ elements) are compared in this study.

    View all citing articles on Scopus
    View full text