This article examines the design and performance of a drill sheet pile wall for a deep excavation in the centre of Copenhagen. Use of a flexible pile wall instead of a stiff concrete diaphragm wall proved very economical and very competitive with other methods.
Introduction Construction of Copenhagen's mini-metro began in early 1997.Built both above and below ground and with a total length of about 20km, the section passing under the city centre consists of two 8km long bored tunnels with other underground constructions such as stations, shafts and crossovers.
Due for completion in 2004, the metro's total cost will be more than $1bn.Three tunnelling methods are being used: TBM (tunnel boring method), NATM (New Austrian Tunnelling Method) and cut and cover. TBM is being used for the main part of the metro, about 7.4km. Cut and cover is used where the tunnel meets the ground.
Project description The pile wall is at Frederiskberg, one of the city centre underground stations (Figure 1) in a densely populated area.The tunnel and station is constructed by cut and cover and the excavation for the project is more than 200m long, 8m to 12m deep and 25m to 30m wide.Figure 2 shows a longitudinal section of the project. To the north side lies a new shopping precinct, Frederiksberg Centre, which is a quite heavy fivestorey building only 0.3m from the pile wall. Vertical and horizontal displacements of the building are restricted to 5mm.
To the south of the project is the Old Frederiksberg station. Built in 1864, it is the oldest railway station in Denmark and had to be protected during the construction of the metro. These two buildings made the excavation a very difficult and challenging foundation engineering project.
The ground surface level is approximately +11.5m.The soil profile includes a fill layer of 0.5m to 1m thick, an upper clay till layer of 4m to 7m, a sand/sand till 2m to 4m thick and a lower clay till layer of 6m to 7m. The clay till is heavily overconsolidated and has a cuvalue of between 150kPa for the upper layer and 300kPa for the lower layer.Limestone occurs at level -4m to -6m.
The modulus of compressibility of the clay till and sand till layers increases with depth and depends on the vertical effective stress. Groundwater level is +2.5m, ie 9m below the ground surface. The design values can be obtained by using the partial factors according to Table 1 .
Drilled sheet-pile HB-wall
Soil conditions combined with the environmental requirements for this densely populated area made excavation design very difficult. The original proposal was for a stiff concrete diaphragm to support the excavation, which would have cost about $5.5M. However, in many places there was no space for a diaphragm wall, especially at the Frederiksberg Centre. One suggestion was to move the old station temporarily and then replace it after construction was completed. This would have cost about $2M.
Stabilator AB proposed a quite flexible and very economical solution: a drilled pile wall, named HB-wall after its inventor Dr Hakan Bredenberg. Each pile consists of a steel tube 194mm in diameter and 6.3mm thick, and a steel HEB-beam. The tube is drilled into the limestone and then the HEB-beam inserted within it and drilled further into the limestone.
The space between the tube and the HEB-beam is filled with concrete and soil between the piles is shotcreted. The pile wall is supported by injected Supa-lina cable anchors with an declination of 30degrees to 35degrees spaced at 2m to 4m.
The method chosen to support the Old Station, designed by Stabilator AB and Skanska Teknick AB, comprised a steel core instead of the HEB-beam to increase the vertical bearing capacity of the piles. The Old Station was then supported by a beam system connected to the foundation wall and anchored to the limestone by vertical GWS anchors. The piles under the building will function both as sheet piles as well as underpinning piles.Figure 3 shows the pile wall with the three rows of anchors and
Figure 4 shows the pile wall beneath the station.
Eurocode 7 Eurocode 7 was used for the design. The ULS (ultimate limit state) and SLS (serviceability limit state) analyses is considered. According to Eurocode 7, two separate calculations generally have to be performed for ULS: one for case B, and another for case C.The partial factors for different calculation cases are shown in Ta b l e 1 .The partial coefficient for deformation, both the modulus of compressibility and its increase with depth, is taken as 1.2.Therefore, for all sections three different cases were analysed according to Eurocode 7: the ULS case B and C, and the SLS with the rare combination load case.
Calculation results from the USL analysis were used for determining the dimensions of the pile wall structure. The soil displacements from the SLS analysis were used for adjusting the possible damage for the existing buildings nearby.
Finite element analysis FEM analysis was carried out with the latest version of Plaxis, Version 7.1 (Finite Element Code for Soil and Rock Plasticity), a program developed by Plaxis BV and Delft University, the Netherlands.
In all six different sections have been calculated. Three FEM analyses have been made for each section.Two most critical sections, the pile wall along the Frederiksberg Centre and the wall below the Old Station at the South side, are described below.
Calculations were carried out by choosing the plane-strain analysis with 15-noded elements.A Mohr-Coulomb model and drained material were used for simulating the soil. The sheet pile wall was modelled by the structural elements. Between the structural elements and the soil elements, interface elements were used. The anchor was modelled by the node-to-node anchor elements, while the grouting body of the anchors was modelled by the geotextile elements together with the interface.
For drained analysis a free water level of +2.5m was assumed.The pile wall is not watertight, so ground water can flow through it.In the analyses all the staged constructions were simulated.
In traditional sheet piling or retaining walls, it is commonly assumed that the soil is active behind the wall and passive in front. Different (active and passive) soil properties were first suggested for simulating a anisotropy problem, which can be a inherent anisotropy or a stress induced anisotropy (see Ladd & Foott 1974). However, it should be noted that with anchor prestressing loads, the classic concept of Rankine active and passive sides is no longer correct. In reality, according to the Rankine concept, the soil at the so-called active side behind a pre-stressed anchor will behave passively.The same shear strength parameters should therefore be used for soil both behind and in front of the wall in FEM analysis.
Design of pile wall along the Frederiksberg Centre
One of the most critical parts of the project is the pile wall along the Frederiksberg Centre, a heavy retail building located only 0.3m from the sheet pile wall. The load from the Frederiksberg centre Table 1: partial safety factors according to Eurocode 7 part 1
Actions Ground properties Case Permanent ariable unfavourable favourable unfavourable tanf¢ c¢ cuquA 1.00 1.50 B 1.35 1.50 C 1.00 1.30 used for the calculations was very large: 6kN/m for USL case B.Despite this large load, the client required the soil displacements under the foundation to be less than 5mm.To meet this, a very high prestressing anchor force was used.
The staged constructions are simulated with the following steps:
l simulate the existing foundation (foundation wall and basement floor) and reset displacements to zero l excavate to the level of the existing foundation +3.5m l install the pile wall and excavate to +2.5m l install the anchor with a prestress l excavate to the final level of the excavation bottom, -0.5m The results show that the displacements of the Frederiksberg Centre foundation in all the construction stages was very small, less than 1mm for horizontal displacements and less than 5mm for settlement (vertical displacements) - see Figure 5. The angular distortion, which is estimated for a structure span of 7m, is about 1/1400. These values are accepted by the client. Results eight months after completion of the excavation, show that the maximum settlement and horizontal displacement of the foundation are both less than 1mm.
In order to compare with the FEM analysis for this section, two traditional calculation methods were used: the analytical Rankine method, and the Line-Polygon method, which is often used in engineering practice. These two methods can not predict the soil movements, only the maximum bending moment in the pile wall and the anchor axial loads. The bending moment obtained from FEM analysis Mmax
= 42kNm/m (case B) and 49kNm/m (case C) are comparable with the values obtained by the traditional methods. Using a reduction factor suggested by Rowe (1952), Rankine gave the maximum bending moment Mmax
= 54 kNm/m (case B) and 55 kNm/m (case C), and the Line-Polygon method,57 kNm/m.
Design of the pile wall under Old Frederiksberg Station
Another difficult part of the project is the pile wall under the Old Station. To avoid moving the station, the pile wall had two functions: to act as a wall and to underpin the structure. The staged constructions were simulated with the following steps:
l simulate the existing foundation and reset displacements to zero l install the pile wall and the tieback anchor l apply the house load and prestress the tieback anchor l excavate to level +8.5m l install and prestress the first anchor l excavate to level +5.0m l install and prestress the second anchor l excavate to level +1.5m l install and prestress the third anchor l excavate to the final level of the excavation bottom, -0.5m, ie 12m deep.
The calculation results show that the maximum bending moment in the pile is 35kNm/m. Soil displacements are quite small: the maximum horizontal displacement is 5mm, while the maximum settlement of the foundation resting at the pile top is less than 1mm.
The project demonstrated that it was possible to use the Stabilator HB pile wall, a very flexible wall, instead of a stiff diaphragm wall in the clay till conditions of Copenhagen. The resulting savings in time and cost were considerable.
Use of the finite element method, Plaxis, makes it possible to design the optimal configuration of anchors and pile wall. In very difficult cases where deformation requirements are critical, FEM is a very effective tool for predicting deformations and displacements.This can be seen especially in the design of the pile wall along the Frederiksberg Centre.
Although the pile wall is quite slender the maximum bending moment is far from reaching the ultimate pile bending resistance.At the wall along the Frederiksberg Centre, the maximum bending moment was approximately 50kNm, while the ultimate pile bending resistance was around 130kNm, considering the combined contribution of the steel tube, the HEB beam and the filling concrete. If the composite interaction is taken into consideration this resistance is even higher.
From full-scale model tests made by Stabilator, an ultimate bending capacity about 160kNm was obtained. This agrees well with the results from a FEM three-dimensional analysis using the 3D FEM program Abaqus.
Ladd CC & Foott R (1974).New design procedure for stability of soft clay.ASCE, GT7, pp 763-767.
Phung Duc Long & Bredenberg, H. (1995).Plaxis - ett effektivtberakningsverktyg for geotekniker.Bygg & Teknik, No 12.
Phung Duc Long (1998). Design of pile wall at Frederiksberg Station, Copenhagen, Section: 3:237 North. Stabilator AB, Gothenburg, December.34 Phung Duc Long (1999). Design of temporary pile wall by Finite Element Method. Project: Frederiksberg, Copenhagen.
Stabilator AB, Gothenburg, March.
Rowe, PW (1952).Anchored sheet-pile walls.Proceedings, Institution of Civil Engineers, Vol.1, London.
Vermeer, PA et al. (1998).Plaxis (Finite Element Code for Soil and Rock Plasticity) Version 7.1 - Users manual.AA Balkema, Rotterdam.