۱- Introduction

۲- Centrifuge model tests

۳- Numerical simulation model

۴- Initial stress conditions

۵- Dynamic simulation results

۶- Discussion

۷- Conclusion

References

Abstract

Numerical simulations of a centrifuge model test of an embankment on a liquefiable foundation layer treated with soil-cement walls are presented. The centrifuge model was tested on a 9-m radius centrifuge and corresponded to a 28 m tall embankment underlain by a 9 m thick saturated loose sand layer. Soil-cement walls were constructed through the loose sand layer over a 30 m long section near the toe of the embankment and covered with a 7.5 m tall berm. The model was shaken with a scaled earthquake motion having peak horizontal base accelerations of 0.05 g, 0.26 g, and 0.54 g in the first, second, and third events, respectively. The latter two shaking events caused liquefaction in the loose sand layer. Crack detectors embedded in the soil-cement walls showed that they developed only minor cracks in the second shaking event, but sheared through their full length in the last shaking event. The results of the centrifuge model test and two-dimensional nonlinear dynamic simulations are compared for the two stronger shaking events using procedures common in engineering practice. The effects of various input parameters and approximations on simulation results are examined. Capabilities and limitations in the two-dimensional simulations of soil-cement wall reinforcement systems, with both liquefaction and soil-cement cracking effects, are discussed. Implications for practice are discussed.

Introduction

Soil-cement grid and wall systems have been used to remediate embankment dams and other civil infrastructure against the effects of earthquake-induced liquefaction in their foundations. Soil-cement treatments have the advantage that they can be constructed in a wide range of soils, including silty soils that can be difficult to treat by densification techniques. A soil-cement grid or wall system is often constructed near the toe of an embankment and covered with an overlying berm to increase confinement and reduce deformations that bypass the treatment zone. An example of this type of configuration is the remediation at the 24-m tall Clemson Upper and Lower Diversion Dams (Wooten and Foreman [23]) as shown in Fig. 1. Other embankment dam remediation projects using soil-cement grid or wall systems in the US include: Sunset North Basin Dam, CA (about 23 m high; Barron et al. [2]); San Pablo Dam, CA (about 44 m high; Kirby et al. [14]); Perris Dam, CA (about 39 m high; Friesen and Balakrishnan [9]), and Chabot Dam, CA (about 30 m high; EBMUD). The seismic performance of soil-cement grids and walls have been studied using three-dimensional (3D) analysis methods (e.g., Fukutake and Ohtsuki [10], Namikawa et al. [16]), but design practices generally rely on two-dimensional (2D) approximations with equivalent composite strengths for the treatment zones (e.g. Wooten and Foreman [23], Barron et al. [2], Kirby et al. [14], Friesen and Balakrishnan [9]). Some common concerns in the design of soil-cement grids for liquefaction remediation include the potential for cracking and brittle failure in the soil-cement elements, the ability of 2D analysis procedures to approximate the 3D response, and the lack of experimental or case history data to validate 2D or 3D numerical analysis methods.