Anda di halaman 1dari 8

EURO:TUN 2009 2nd International Conference on Computational Methods in Tunnelling Ruhr University Bochum, 9-11 September 2009 Aedificatio

Publishers, 1-4

Plaxis-GiD Modelling of Tunnel-Pile Interaction


Siew-wei Lee1, William Cheang2, Wendy Swolfs3 and Ronald Brinkgreve3 1 Geotechnical Consulting Group (Asia) Ltd, Hong Kong 2 Plaxis (Asia), Singapore 3 Plaxis BV, the Netherlands Abstract Construction of tunnels adjacent to/beneath existing piled foundations poses challenges to geotechnical engineers to realistically estimate pile deformations and stresses induced by tunnelling. Use of design charts established using closed-form solutions/boundary element methods may only be suitable as a first approximation because of assumptions made on ground conditions and geometries for tunnels and piles. 3D numerical modelling can better model the real soil and structure conditions encountered in the field. This paper uses a new finite element programme, Plaxis-GiD to model a tunnelling case history in Singapore. Contract 704 involved construction of twin tunnels in close proximity to 1.2 m diameter piled foundations for a viaduct bridge. The 3D analysis has considered non-linear soil stiffness from small strains. The Stress Relaxation method has been used to model excavations for the respective Southbound tunnel and Northbound tunnel at steady state settlement conditions. Interface elements are specified down the piles and around the perimeter of the pile cap. The predicted tunnelling-induced ground movements and pile axial forces and bending moments are in reasonable agreements with the measurements. It is demonstrated that Plaxis-GiD can model complicated soil behaviours and tunnel-pile interaction problems.
Keywords: Numerical modelling, measurement, prediction, pile, tunnel

Siew-wei Lee, William Cheang, Wendy Swolfs and Ronald Brinkgreve

INTRODUCTION

Construction for new developments is often carried out in built-up areas due to land scarcity. New tunnels have been constructed beneath/adjacent to piled foundations supporting existing buildings/infrastructures. The effect of tunnelling on exiting piles in homogeneous soil conditions has been studied using closed-form solutions, boundary element methods, centrifuge and numerical modelling. However, field observations on movements and stress changes in piles in response to tunnelling are scarce, due to difficulties associated with instrumenting the existing piles. Pang [2] presents a unique case study of tunnelling induced axial forces and bending moments in well-instrumented piles for Contract 704, North-East Line in Singapore. This experience has shed light on the potential for piles stressed in excess of their structural capacity due to ground movements induced by tunnelling. This paper models the tunnelling experience at Contract 704 using a new threedimensional (3D) finite element (FE) programme, Plaxis-GiD. The twin tunnels, individual piles and pile cap have been modelled, and the soils are represented by a constitutive model incorporating non-linear soil stiffness from small strains. Predictions are compared to the field measurements. It is demonstrated that using advances in 3D FE modelling, engineers could better estimate the response of piled foundations to tunnelling. 2 BACKGROUND INFORMATION FOR CONTRACT 704

Contract 704 involved the construction of 2.5 km long twin tunnels of 6.5 m diameter using two Earth Pressure Balance machines (EPBM) from Woodleigh Station via Serangoon Station to Kovan Station, and a 1.9 km long dual-lane viaduct bridge (Pang [2]). Figure 1 shows the geological profile for Contract 704 and a schematic diagram of the Southbound (S/B) and Northbound (N/B) tunnels relative to the 22 pile group for Pier 20 investigated herein. The 1.2 m diameter, 60.5 m long piles were instrumented and installed between the twin tunnels before tunnelling commenced. The edges of Pile P1 (front pile) and Pile P2 (rear pile) are located at 1.6 m and 3.3 m respectively from the respective S/B and N/B tunnel extrados. The pile cap size is 5.3 5.3 1.5 m. The S/B tunnel was excavated first and followed by the N/B tunnel which lagged behind by 200 to 300 m. At Pier 20, the ground conditions comprise completely weathered Bukit Timah granite material (Grade VI material) formed during the lower to middle Triassic

Plaxis-GiD Modelling of Tunnel-Pile Interaction

period (Pang [2]). The material is locally known as the G4 material and is predominantly reddish brown, sandy silty clay. Within the G4 material, further layering is made according to the SPT-N value.

G4a (N<15) -16m G4b (15<N<30) -30m G4c (30<N<50) -44m G4d (50<N<100)

Strain gauges

Not to scale

Figure 1:

Contract 704 geological profile and tunnelling geometry near Pier 20 (after Pang [2]).

PLAXIS-GID MODELLING OF TUNNELLING

Figure 2 shows the 3D model set-up for modelling the Contract 704 tunnelling near Pier 20. This model takes symmetry about the middle of the 22 pile group, and there are 24,000 10-node tetrahedral elements. Interface elements have been specified down the 60.5 m long piles and around the perimeter of the pile cap. The Stress Relaxation Method has been adopted to model the tunnel excavation, where an internal support pressure has been applied on the inner perimeter of the excavated tunnel to achieve the measured ground loss ratios (VL) of 1.38% and 1.67% for the S/B and N/B tunnels respectively. For modelling of tunnelling for Contract 704, the progressive advance of the tunnel face has not been modelled. (Note that the programme can model advance of tunnelling). The main objective is to model the situations when the respective S/B and N/B tunnel faces have well passed Pier 20, achieving steady-state settlement profiles (i.e. plane strain condition) above the two tunnels. In the analysis, the entire length of the S/B tunnel modelled has been excavated all at once and followed by the N/B tunnel in a similar manner. This is a simplification of the real 3D tunnelling, hence the pile behaviour in the direction parallel to the tunnel alignment cannot be assessed.

Siew-wei Lee, William Cheang, Wendy Swolfs and Ronald Brinkgreve

0 to -16m: G4a -16 to -30m: G4b -30 to -44m: G4c Below -44m: G4d

Half pile cap

Interfaces down piles & around cap perimeter

Tunnels N/B tunnel S/B tunnel

Symmetric plane Size 1207520m

P2 Piles

P1

Figure 2:

Plaxis-GiD 3D model for modelling tunnelling near Pier 20.

The weathered Bukit Timah granite layers (G4 materials) have been modelled using the Hardening Soil-Small (HSsmall) model in Plaxis-GiD. Information about the HSsmall model and associated input parameters can be found in Benz [1]. Pang [2] uses the Strain Dependent Modified Cam Clay Model (SDMCC) implemented as a subroutine in programme Abaqus to consider the non-linear stiffness from small strains for the G4 materials, with the model input parameters calibrated against the results of pressuremeter tests in the G4 materials and field/laboratory tests on completely weathered granite samples in Hong Kong. Figure 3 shows that the small strain non-linear stiffness curves predicted by the SDMCC and HSsmall models are very similar. Table 1 presents the input parameters for the HSsmall model. All G4 materials have been modelled as a drained material, considering that their permeability values are relatively high at 110-7 m/s. The groundwater table is specified at 3 metre below ground level (mbgl). For initial stress equilibrium, the effective horizontal stress is taken to be similar to the effective vertical stress (Pang [2]). The piles and pile cap have been modelled by continuum (or volume) elements using the linear elastic constitutive model with a Youngs modulus (E) of 28 GPa and a Poissons ratio () of 0.2. The tunnel linings have been modelled using Plate structural elements with an E of 28 GPa, of 0.2 and a thickness of 0.25 m.

Plaxis-GiD Modelling of Tunnel-Pile Interaction

Table 1: Soil kN/m G4a G4b G4c G4d 18 19 20 20


3

Table 1: Input parameters for the HSsmall model. E50ref MPa 15 20 35 50 Eoedref MPa 15 20 35 50 Eurref MPa 60 80 140 200 m [-] 0.8 0.8 0.8 0.8 0.7 [-] 2E-5 2E-5 2E-5 2E-5 G0 MPa 98 130 228 325 pref kPa 100 100 100 100 K0nc [-] 0.53 0.5 0.5 0.5 [-] 0.2 0.2 0.2 0.2 c' kPa 5 5 5 5 ' 28 30 30 30

1 000 SDM CC 800 600 400 200 0 0.0001 0.001 0.01 Shear strain (%) 0.1 1 Hssmall_G4a Hssmall_G4b 0.000 -0.005 -0.01 0 -0.01 5 -0.020 -0.025 -0.030 -0.035 -0.040

Distance fro m pile cap centreline (m) -60 -40 -20 0 20 40 60

S/B Greenfield (M ) measured S/B Greenfield (P ) predicted

Figure 3:

HSsmall non-linear stiffness curves. Figure 4: Settlements after S/B tunnel passage.

Figures 4 and 5 compare the predicted ground surface settlements with the measured settlements at the Pier 20 section when the S/B tunnel and both the S/B and N/B tunnels (S/B + N/B) respectively had well passed the pier. The predicted pattern and magnitude of the ground surface settlements are close to the measurements. This demonstrates the applicability of the HSsmall model in replicating the gradients of the measured surface settlement profiles. Figure 6 compares the predicted subsurface horizontal displacements with the measurements at a greenfield vertical section corresponding to the Pile P1 location when the S/B tunnel had well passed Pier 20. A reasonable agreement is achieved between the predicted and measured pattern and magnitude of the subsurface horizontal displacements.

Siew-wei Lee, William Cheang, Wendy Swolfs and Ronald Brinkgreve

-60 0.000 -0.005 -0.01 0 -0.01 5 -0.020 -0.025 -0.030 -0.035 -0.040

Distance fro m pile cap centreline (m) -40 -20 0 20 40

60

0.000 0 -1 0 -20 -30 -40 -50

Subsurface ho rizo ntal disp. (m) 0.005 0.01 0

0.01 5

S/B Greenfiled (M ) S/B Greenfield (P )

S/B + N/B (pile cap) (M ) S/B + N/B (pile cap) (P ) S/B + N/B (Greenfield) (P )

-60 -70

Figure 5:

Settlements after S/B + N/B passage.

Figure 6: Subsurface horizontal displacements.

Figure 7 compares the predicted axial forces in Pile 1 (front) and Pile 2 (rear) with the measurements when the S/B tunnel and the S/B + N/B tunnels had well passed Pier 20. For Pile 1, the predictions generally match the measured pile axial forces above the tunnel axis level. Note that the measured maximum pile axial force of 5100 kN occurred below the tunnel invert level, which was not predicted by the analysis. For Pile 2, the predictions generally match the pattern of distribution of the pile axial forces, but under predict their magnitude.
0 0 -1 0 -20 -30 -40 -50 -60 -70 SB (M ) SB +NB (M ) S/B (P ) 1 000 A xial fo rce (kN) 2000 3000 4000 5000 6000 0 0 1 000 A xial fo rce (kN) 2000 3000 4000 5000 6000

P2

-1 0 -20 -30 -40 -50 -60

P1
Crown Invert

SB (M ) SB +NB (M ) S/B (P ) S/B +N/B (P )

S/B +N/B (P ) -70

Figure 7:

Comparison of pile axial forces.

Figure 8 compares the predicted bending moments in Pile 1 (front) and Pile 2 (rear) with the measurements when the S/B tunnel and the S/B + N/B tunnels had well

Plaxis-GiD Modelling of Tunnel-Pile Interaction

passed Pier 20. For Pile 1, the predictions match well the pattern and magnitude of the measured pile bending moments. For Pile 2, the predictions reasonably match the measured pile moments due to the S/B tunnel excavation. The predictions however do not match the measured moments after the S/B + N/B tunnels had well passed the pier, comparing the measured maximum -165 kNm to the predicted +175 kNm.
B ending mo ment, M (kNm) -400 0 -1 0 -20 -30 -40 SB (M ) -50 -60 -70 SB +NB (M ) S/B (P ) S/B +N/B (P ) -200 0 200 400 600 800

B ending mo ment, M (kNm) -400 0 -200 0 200 400 600 800

P2
-1 0

P1
-20 -30 -40 SB (M ) -50 -60 -70 SB +NB (M ) S/B (P ) S/B +N/B (P )

Crown Invert

Figure 8:

Comparison of pile bending moments.

M icro strain [-] 0 0 -5 -1 0 -1 5 -20 -25 -30 -50 -1 00 -1 50 -200

Y2 N/B X1 Pile Y1 X2 S/B

Sudden increase in displacement

Tunnel axis X1 X2 Y1 Y2

Figure 9:

Strain gauges readings.

Figure 10: Tunnel advance vs. soil horizontal displacements.

It is noted that the predictions on the measured ground movements are better than those on the measured pile axial forces and bending moments. Pang [2] highlights that the readings from a set of four strain gauges installed at a particular pile elevation

Siew-wei Lee, William Cheang, Wendy Swolfs and Ronald Brinkgreve

showed large variations amongst themselves, especially those near the tunnel level. Figure 9 shows an example of strain readings from the strain gauges. Note that the pile forces and moments are back-calculated from the measured strains, which might contain some inconsistencies. Another reason for the deviations between the predicted and measured pile forces and moments may be due to the omission of modelling progressive advance of the tunnel face. Figure 10 shows an example of a sudden increase in the measured soil horizontal displacements near Pile 2 when the N/B tunnel face just passed the pile. This resulted in a sharp increase of the pile bending moment in the opposite direction (see measured P2 moment -165 kNm in Figure 8). 4 CONCLUSION

A new FE programme, Plaxis-GiD, has been used to model the effect of tunnelling on adjacent piled foundations using a tunnelling case study in Singapore. The HSsmall model has been used to consider non-linear soil stiffness from small strains for the weathered Bukit Timah granite materials. The predictions show good agreements with the measured ground surface settlements and subsurface horizontal displacements. The predicted pile axial forces and bending moments show some deviations from the measurements, probably due to large variations in the strain gauge readings and to the influence of progressive advance of tunnelling. This paper has demonstrated the applicability of Plaxis-GiD to model complicated tunnel-pile interaction problems. Most of the existing tunnel-pile design charts have been developed using closed-form solutions and/or boundary element methods, based on assumptions of homogeneous ground conditions, particular sets of pile and tunnel geometries and their relative distances. Care should be exercised when using these design charts on real tunnelling problems where the ground conditions and pile and tunnel geometries can be significantly different. 3D FE modelling offers an alternative design tool to engineers to better understand and estimate the pile behaviour and ground movements in response to adjacent tunnelling. REFERENCES [1] Benz, T. (2007). Small-strain stiffness of soils and its numerical consequences. PhD thesis, University of Stuttgart. [2] Pang, C. H. (2006). The effects of tunnel construction on nearby pile foundation. PhD thesis, National University of Singapore.

Anda mungkin juga menyukai