Challenges of quick clay excavation in urban area with sloping ground

Transcription

1 Challenges of quick clay excavation in urban area with sloping ground R. M. Nalbant 1 Multiconsult AS, Oslo, Norway ABSTRACT The paper presents the overall stability evaluation for a deep excavation in quick clay supported by flexible retaining structure (sheet pile wall) and ground improvement by lime-cement columns and vertical drains. The soil structure interaction analysis is performed using the finite element program Plaxis. The challenge of modelling the sloping layers with shear strength increasing linearly with depth is managed by using a special vertical clusters technique. Main conclusions from Plaxis analyses are as follows: Mohr-Coulomb model can be used to simulate pre-peak behavior of quick clay and the FE method can detect most critical failure mechanisms. The working sequences are monitored by lateral displacement (inclinometers), settlement and porewater pressure measurements. Unfortunately, due to some site problems the quality of measurements is questionable. Despite some technical problems during the excavation work, the bottom of the excavation is now successfully closed by a concrete bottom slab and the structural work is in progress. Keywords: sheetpile wall, quick clay, stability, numerical analysis, monitoring 1 INTRODUCTION This paper presents an overall stability evaluation for a deep excavation in quick clay. The excavation is supported by a flexible retaining structure (multi-supported sheet pile wall), and the ground is improved by means of lime-cement columns and vertical drains. The soil structure interaction analysis is performed using the finite element program Plaxis. The purpose of the analyses is to document that there is satisfactory stability of the area during and after excavation together with sheet pile wall design. 2 PROJECT DESCRIPTION 2.1 General The excavation is located in Drammen, Norway, some 4 kilometers southwest of Oslo. The 4-storey office building with 1-2 basement floors is founded directly on a concrete bottom slab. The terrain inclination is relatively gentle ( 1:16 in average) resulting in approximately 7 m excavation upstream and 4 m excavation downstream, see Figure Geotechnical challenges of excavation in quick clay The geotechnical challenges are as follows: 1 Multiconsult AS, Nedre Skøyen vei 2, 276 Oslo, Norway. roy.michel.nalbant@multiconsult.no

2 - Stability of the building pit during and after excavation; - Design of ground improvement by means of lime-cement columns ; - Design of sheet pile wall. The paper shows how these challenges were accounted for using Plaxis finite element analyses. Calculation phases 2, 7 and 8 are shown in Figure 3, 4 and 5, respectively. Table 1. Calculation phases in Plaxis together with a description Calculation Description Phase In-situ stresses (gravity loading). 1 Installation of sheet pile wall. Installation of prestressed upper 2 anchor(s) in local trenches. 3 1 m excavation. 4 Installation of lime-cement columns. 5 Further 1 m excavation. Local excavation prior to installation 6 of lower anchor(s). Installation of prestressed lower 7 anchor(s). 8 Final excavation. Figure 1. Sketch of the 4-storey office building with 1-2 basement floors. 3 PLAXIS MODEL The Plaxis model is shown in Figure 2. Figure 3. Installation of prestressed upper anchor(s) (node-tonode anchor) to bedrock. Installed in local trenches. Note: the sheet pile wall is fixed at bedrock with tip-bolts. The anchors at the tip have to be included because the interface-element deactivates the boundary condition. Figure 2. The Plaxis model with 8 15-node triangular elements with 12 Gauss integration points. The total horizontal dimension is approx. 3 m. The upper 2 m, consisting of fill compound, is modelled with the Hardening Soil model. The other layers consisting of 3 m silty clay over quick clay and some stiffer clay over the bedrock are modelled with the Mohr-Coulomb model. The depth to bedrock in the sheet pile wall center line is 3 m. Table 1 shows an overview of the calculation phases in Plaxis together with a description. Figure 4. Installation of prestressed lower anchor(s) (node-tonode anchor) to bedrock. Note: the lime-cement columns in the building pit are modelled with increased strength, see Table 3.

3 Figure 5. Final excavation. Note: The unloading stiffness in excavation phases are modelled as 3 times the loading stiffness. stabilized material the undrained shear strength and stiffness are interpreted based on laboratory tests. Linear increase in strength and stiffness are modelled in Plaxis using c ref /c incr and E ref /E incr together with a reference level y ref. In order to be able to model linear increase in strength and stiffness for the sloping layers, the layers are vertically divided in an appropriate number of clusters with the corresponding reference levels; this is illustrated in Figure 7. It is noted that a negative value of c incr and E incr is not possible in Plaxis and will be read as (zero). This can however be solved by dividing the layer in sublayers with constant c and E. Figure 6. Interpretation of undrained shear strength from CPTU tests. 4 SOIL PARAMETERS The characteristic undrained shear strength in the soil profile is interpreted based on results from laboratory tests and cone penetration tests (CPTU). The interpretation is presented in Figure 6. The stiffness (E 5 ) is assumed 45 times the direct undrained shear strength. For lime-cement Figure 7. Illustration of linear increase of strength and stiffness in Plaxis for sloping layer, without and with vertical clusters. The undrained behavior of the silty clay, quick clay and stiff clay layers is modelled using the drained Mohr-Coloumb soil model with undrained parameters, see Table 2. Input parameters for the lime-cement stabilized materials are shown in Table 3.

4 Table 2. Input parameters for drained Mohr-Coulomb soil model used to model the undrained behavior of soil. Parame Silty Quick Unit ter Clay Clay Clay γ (kn/m 3 ) E ref E incr ν c ref c incr ϕ ψ T -Strength R inter (-) (-) Stiffness multiplied by 3 in excavation phases for underlying soil clusters to model unloading stiffness with the Mohr-Coulomb model. 3D-effects are to some extent included by increasing the undrained shear strength with 5 % based on corresponding increase of stability number for 3D compared to 2D conditions. Figure 8. Deformed mesh after final excavation. Maximum deflection 88 mm. Scaled 5 times. Table 3. Input parameters for drained Mohr-Coulomb soil model used to model the undrained behavior of lime-cement. Parametecement cement Lime- Lime- Unit γ (kn/m 3 ) E ref 15 3 E incr ν c ref c incr ϕ ψ T -Strength (-) (-) R inter Represents the lime-cement improved soil volume above final excavation level (half mixture of lime-cement for digging ability). Represents lime-cement improved soil volume below final excavation (full mixture of lime-cement). a) Max. 6 cm b) Max. 259 knm/m c) Max. 199 kn/m 5 RESULTS Figure 9. Sheet pile wall displacements (a), and cross section forces: moment (b) and shear force (c) after final excavation. The deformed mesh after final excavation is shown in Figure 8. Figure 9 shows the sheet pile wall deformation, moment and shear force after final excavation. Figure 1 shows the shear strains (total strains shear shadings). Figure 1. Shear strains (total strains shear shadings) after final excavation. Maximum value 1 %.

5 Figure 11 shows the calculated safety factors and failure mechanisms on which the calculation of safety are based, prior to excavation and after final excavation. A simplified hand calculation for bottom heave stability based on local failure mechanism shows a safety factor of F = 1.4. Use of Plaxis helps therefore finding the most critical bottom heave failure mechanism showing a safety factor of F = The anchors and sheet pile wall are modelled as elasto-plastic materials to avoid unrealistic wall forces when calculating the safety factors. It is further noted that the safety factors calculated using s ud are conservative since anisotropic shear strength (s ua over the longest active part of the sliding surface and s up + s u,lime-cement column over the shortest part) will result in higher safety factors. Figure 11. Failure mechanisms a) prior to excavation and b) after final excavation. Figure 12 shows a cross section of the terrain deformations over 4 m long distance behind the sheet pile wall after final excavation. The relative simple Mohr-Coulomb (M-C) soil model is used in Plaxis to model strength and stiffness of quick clay. The M-C soil model is a linear elastic, perfectly plastic model, and an approach to verify whether the degree of soil mobilization will result in significant strain softening effects or not, is to check the strains during the excavation sequences. Strains at which quick clay shows significant strain softening will vary, but shear strains in the order of % should be a fair estimate. A more reliable determination of this shear strain can be obtained from laboratory tests on high quality undisturbed samples, such as the ones from the Sherbrook sampler. The linear increase in strength and stiffness is modelled by vertical division with the appropriate reference level. The dimensions of the division depend on the slope of the ground. In this case the clusters are mainly divided vertically for each 1 m. For the time being limited adequate soil model together with an appropriate numerical solution exists for quick clay modeling. However, for engineering purposes the relative simple M-C soil model may be used bearing in mind the limitations this model involves together with checks of the calculated strains. Main conclusions from the Plaxis analyses are as follows: Mohr-Coulomb model can be used to simulate the pre-peak behavior of quick clay and the FE method can detect most critical failure mechanisms. ACKNOWLEDGEMENT Figure 12. Terrain deformations over 4 m long distance behind the sheet pile wall after final excavation. 6 DISCUSSION AND CONCLUSIONS Quick clay shows usually a strong strain softening behavior, as shown in Figure 5 in Bjerrum state-of-the-art Report [2], which must be included in stability analyses to avoid unrealistic soil mobilization distribution along a critical slip surface. The author wish to extend special thanks to Corneliu Athanasiu, Anders Bye, Jan Finstad and Knut Espedal, colleagues at Multiconsult AS, for their contribution both in the project and preparation of this article. REFERENCES [1] Brinkgreve, R.B.J, Broere, W., Waterman, D. (28) Plaxis 2D - Version 9., Delft University of Technology & Plaxis b.v., The Netherlands.

6 [2] Bjerrum, L., (1973) Problems of Soil Mechanics and Construction on Soft Clays, State-of-the-Art Report to Session IV, 8 th International Conference on Soil Mechanics and Foundation Engineering, Moscow.