You are on page 1of 13

Case Study: Negative Skin Friction Development on Large Pile Groups

for the New Wembley Stadium

Mei T. Cheong, Mott MacDonald, UK

1 Introduction

There have been several studies of the influence of negative skin friction (NSF) on pile
behaviour (Shibata et al., 1982; Jeong et al., 2004; Lee et al., 2004; Chellis, 1996; Garlanger
et al., 1973). Most of these studies have been either theoretical or based on the behaviour of
small scale models. Observations of the development of NSF on full scale pile groups are
extremely rare.

NSF occurs when the soil next to the pile settles more than the pile. The settling soil induces
additional axial force (drag force) on the pile, resulting in additional pile settlement (drag
down). There are two common design situations:

(i). there is a clear change in strength and stiffness with depth, e.g. peat over weak rock,
the thickness of the soil layer which will apply NSF to the pile is straightforward to
assess. The main uncertainty is the magnitude of NSF.
(ii). there is a gradual increase in strength and stiffness with depth. The thickness of soil
layer which applies NSF is not obvious. Therefore the magnitude of NSF and the
overall settlement of the “floating” piles becomes a complex soil-structure interaction
problem.

This paper considers the more complex issues associated with (ii) above. Observations of the
effects of NSF on a large pile group are summarised, based on monitoring over a two year
period. Back analysis of the observed pile group behaviour is discussed based on finite
element analysis. Finally the results of two different simplified methods of analysis are
compared, and the practical implications are summarised.

2 Negative Skin Friction on Large Pile Group

The development of the new Wembley Stadium includes a large pile group for the eastern
base of the stadium’s iconic arch structure to be constructed beneath 9.0m thick of clay fill. In
plan, dimensions of the clay fill are approximately 80m x 100m. The consolidation of the clay
fill and the underlying London Clay has led to the development of NSF on this large pile
group.
Long-term monitoring of ground movements beneath the base of the clay fill and the pile cap
movements of the eastern arch base were undertaken over a period of two years following
construction of clay fill. The location of these monitoring points is shown in Figure 1. This
monitoring data has been used to carry out back analyses on the performance of the large pile
group.

S0 Fill
X ME4R

Eastern Arch Base

Key: X -Approximate location of monitoring points

Figure 1: Location of monitoring point

3 Ground and Groundwater Conditions

The new Wembley Stadium is located in North London and the geology at the site comprised
London Clay overlying the Lambeth Group, with Chalk at depth as shown in Figure 2. The
thickness of London Clay varies between 30m and 40m. The unweathered London Clay
becomes siltier and sandier at about 21mOD, marking a change from the upper lithological
unit B to the lower unit A (King, 1981). An important feature at this site is that the depth of
weathering, at between 8m and 10m, is towards the upper end of the range observed for
London Clay sites (Chandler, 2000).

The water table level was identified as approximately 2.5m below the top of the London Clay
surface, based on piezometer monitoring. The ground conditions and geotechnical design
parameters for the stadium were discussed by O’Brien et al. (2004). A schematic layout of the
geometry of the large pile group is also shown in Figure 2. The eastern arch base is supported
by nineteen 1500mm diameter piles with lengths of 27m.

52.5 19.5m

Fill Eastern arch


43.5 base 41mOD
Upper Weathered
Elevation, mOD

37.5 London Clay

32.5 Lower Weathered


London Clay

Upper Unweathered
London Clay
21.0
Lower Unweathered 16.5mOD
13.0 London Clay

Lambeth Group
5.0

Figure 2: Schematic of the site geology and geometry of the eastern arch base and clay fill

4 Finite Element Modelling

Finite element modelling of the eastern arch base was carried out in 2D axisymmetrical
conditions using the software PLAXIS. As this research focuses on the global behaviour of
the pile group rather than individual piles within the group, this 3D problem can be simplified
to a 2D model (Lee et al., 2002).

An elastic model was used for the piles and a Mohr-Coulomb model for the London Clay and
Lambeth Group. The large pile group was modelled using the equivalent pier method
proposed by Poulos (2001). The input soil parameters are summarised in Table 1.
Table 1: Input parameters
Soil Layer Layer Unit Horizontal & Poisson’s Young’s Cohesion, Friction Interface
thickness weight, Vertical Ratio, υ Modulus, E c (kPa) angle, φ Parameter
2
(m) γsat Permeability, (MN/m ) (degrees) R
3
(kN/m ) k (m/s)
z=depth
from top of
London Clay
Clay Fill 9 19 10
-9
0.25 10+0.6z 1 20 1
Upper 0.85
Weathered -8
6 20 10 0.15 5+3z 6+0.2z 20
London
Clay
Lower 0.85
Weathered -9
5 20 10 0.15 25+5z 7+0.6z 20
London
Clay
Upper 0.9
Unweather -10
11.5 20 10 0.15 50+13.6z 10+0.4z 22
ed London
Clay
Lower 0.8
Unweather -9
8 20 10 0.15 220+26z 15+0.4z 22
ed London
Clay
Lambeth -8 1
8 20.5 10 0.2 400+80z 26+0.4z 22
Group

Note: (i) Interface factor, R α fn [Gsoil, c, tan φ soil] where G=shear modulus, c=cohesion, φ =friction
angle.
(ii) The soil stiffness is assumed linear elastic – perfectly plastic, however the variation of the
‘elastic’ modulus with depth is based on the non-linear small-strain stiffness characteristics of
the London Clay.

5 Soil Model Calibration with S0 Fill Monitoring Data

Initial analysis was carried out to calibrate the soil model with the monitoring data obtained
from the rod extensometer (MER4) located beneath the base of the clay fill. Readings were
taken at six different levels ( 0m, 6m, 12m, 18m, 32m and 47m) beneath the base of the fill.
The clay fill was placed between March and August 2003.

Readings show that settlement was rapid within the first year of completion of the fill (Figure
3). In order to match this initial rapid settlement, the upper Weathered London Clay was
modelled to have a relatively high permeability k=10-8m/s. This is thought to be due to the
influence of weathering which is likely to have led to a more open and porous soil structure
(note: permeability, k α fn[e] where e= void ratio, (Vaughan, 1983)).The PLAXIS back
analysis is comparable with the data recorded from ME4R (Figure 3). The time taken to
complete 95% of the primary consolidation is estimated to be about four to five years.
Clay Fill Area Backanalysis with Extensometer Data [ME4R]
Time, days
10 100 1000 10000
0
10
Start of monitoring
20 26 April '03
30 ME4R below base of fill (bbf)
40 ME4R (6m bbf)
ME4R (12m bbf)
Settlement, mm

50 ME4R (18m bbf) A


ME4R (32m bbf)
60 ME4R (47m bbf)
PLAXIS S0 Fill bbf
70
PLAXIS S0 Fill 6m bbf
80 PLAXIS S0 Fill 12m bbf
PLAXIS S0 Fill 18m bbf Primary Consolidation
90 PLAXIS S0 Fill 32m bbf t = 19 years
PLAXIS S0 Fill 47m bbf End of fill p=~120mm
100 B
NLS Total Displacement bbf 03 Aug '03
110 NLS 6m bbf
NLS 12m bbf
120 NLS 18m bbf Start of arch End of arch lift Current time
NLS 32m bbf lift 21 May 04 26 June 04 14 Feb 07
130

Figure 3: PLAXIS simulation of clay fill settlement compared with monitoring data

Clay Fill Area


Settlement with Depth

50

40

30
Elevation, mOD

20

10

-10
0 20 40 60 80
Settlement, mm

Extensometer [ME4R] at Point A


Plaxis Clay Fill at Point A
PLAXIS Single Pile
PLAXIS Pile Group

Figure 4: Settlement with depth from monitoring data and PLAXIS at Point A
Figure 4 shows the settlement at depth beneath the base of the clay fill approximately 300
days after completion of the clay fill construction. Comparable settlement was computed from
the PLAXIS back analysis. It is noted that more than half the settlement occurred within the
top 5m of the London Clay.

6 Pile Group Calibration with Eastern Arch Base Monitoring

The second stage of analysis involved calibrating the equivalent pier within the numerical
model. The average axial load (dead and live) acting on the eastern arch base is
approximately 4MN per pile. A single pile calibration for loads under 5MN was carried out
initially against single pile test data (Figure 5). Pile test data was also compared with an
estimate using the Cemset single pile settlement prediction method proposed by Fleming
(1992). The single pile response was implemented into the numerical model. The calibration
was important to obtain appropriate pile/soil interface parameters (Table 1).

12.0

10.0

8.0
Load (MN)

6.0

4.0
Pile Test data

2.0 PLAXIS Single pile

PLAXIS Pile Group


0.0
0 10 20 30 40 50
Displacement (mm)

Figure 5: Single pile calibration

Precise level monitoring of the pile cap movement of the eastern arch base are presented in
Figure 6, for the period between May 2004 and March 2006. Approximately 300 days after
the lifting of the arch, there was a sudden increment in the vertical displacement of the pile
cap. This was due to the monitoring point being disturbed, and therefore the data was
rezeroed from that point as shown below. It is noted that pile cap settlement continued to
progress after completion of the arch lift, indicating the occurrence of drag down.
Long term Vertical settlement at Eastern Arch Base

10 20
End of arch lift Monitoring data
26 June 04
Rezeroed monitoring data
PLAXIS Pile Group-Pilecap settlement 0
5
Vertical displacement, mm

PLAXIS Single Pile-Pilecap settlement


Load on arch
-20
0

Load, MN
-40

-5
-60

-10
-80
Last monitoring point
14 Mar 06
-15 -100
0 100 200 300 400 500 600 700 800 900
Time, days

Figure 6: Pile cap settlement monitoring compared to PLAXIS prediction


Note: Only incremental settlement after May 2004 is recorded

The load sequence applied on the equivalent pier in PLAXIS is summarised in Table 2.
During the monitoring period the pile cap settled approximately 5-7mm, compared with an
equivalent pier prediction of about 5mm for the same time period.

Table 2: Time line of construction activity


Construction phases Load Component Time Comments
Piles completed Structural vertical 25 Nov 2002 Weight of piles applied
Arch base completed dead load 16 Dec 2002 Weight of pile cap
applied
Start of clay fill Earth structure 23 Mar 2003 Fill weight applied
Start of ME4R monitoring vertical dead load 26 Apr 2003 incrementally
Clay fill completed 03 Aug 2003
Start of arch lift Structural vertical 21 May 2004 Weight of arch applied
Arch lift completed load of arch 26 Jun 2004 incrementally
Note: In the axisymmetric analysis only vertical loads are applied; actual loads included horizontal,
moment and torsional loads.

7 ‘Settling Zone’ within the Eastern Arch Base

Figure 7 shows that incremental settlement at depths between 6m and 10m beneath the base
of fill are similar to the pile cap settlement. The total settlement, at depths between 6m and
10m beneath base of fill, are between 40mm and 20mm respectively. Ignoring soil structure
interaction effects is only slightly conservative (settlements about 5mm less).

Long term Vertical settlement at Eastern Arch Base

10 20
End of arch lift Monitoring data
26 June 04 Rezeroed monitoring data
PLAXIS Pile Group -fill settlement 6m bbf 0
5
Start of arch lift PLAXIS Pile Group-fill settlement 10m bbf
Vertical displacement, mm

21 May 04 PLAXIS Single Pile-fill settlement 6m bbf


-20
PLAXIS Single Pile - fill settlement 10m bbf
0

Load, MN
Load on arch
-40

-5
-60

-10
-80
Last monitoring point
14 March 06
-15 -100
0 100 200 300 400 500 600 700 800 900
Time, days

Figure 7: Soil settlement compared to pile cap monitoring


Note: Only incremental settlement after May 2004 is recorded

A neutral plane exists at depth where the soil settlement and pile settlement are equal and
where the drag force can be computed (Fellenius, 1998). The PLAXIS back analysis shows
the neutral plane for the large pile group to be an average of 8m beneath the base of the fill,
with a likely range varying between 6m and 10m.

8 Design Implications

Sections 4 to 7 has presented the behaviour of a large pile group observed from two years of
monitoring data and also the back analysis carried out using PLAXIS. The subsequent
sections compare two alternative simplified design methods.

8.1 Conventional Design (Linear Elastic/Total Stress)

This section discusses a conventional design approach which could be used to predict the drag
force for a single pile 27m long 1500mm diameter in London Clay. The original ground
surface is at 43.5mOD. The software VDisp, which uses linear elastic theory, is used to
predict the soil settlement under a load equivalent to 9m of clay fill. The input parameters are
presented in Table 3.
Table 3: Undrained shear strength and Young’s Modulus profiles based on U100 testing
Elevation, Thickness, Undrained shear strength U100, Young’s Modulus,
(mOD) (m) su (kPa) E’=200su, (MPa)
0-4m, su=60+17z
>4m, su=115+3.3z

z = depth from top of London


Clay
43.5 0 60 12
39.5 4 129 26
16.5 27 204 41
5 38.5 242 48
Note: E’=200su is a commonly used empirical correlation, refer to Stroud (1989)

Assuming the neutral plane is close to the location at which shaft capacity above and below
the neutral plane is equal (the neutral plane could not be deeper for a ‘stable’ pile group), then
the pile group settlement is close to 100mm (Figure 8); NSF develops over a length of about
15m and NSF is about 4,500 to 5,000 kN (based on τ=α su, where α=0.5, su=undrained shear
strength, kPa). If a shallower and intuitively more reasonable depth of NSF was assumed then
pile group settlement would approach 150mm.

8.2 Alternative Design Solution (Non-linear Elastic/ Effective Stress)

An alternative method for predicting soil settlement is a Non-Linear Elastic method proposed
by O’Brien et al. (2001) for over-consolidated clay. The input parameters are shown in Table
4 and Table 5.

Table 4: Input parameters for NLS


Elevation, mOD Depth, m Initial Mean Effective Young’s Modulus, E’ at
Stress, p’ (kPa) 0.1% strain (MPa)
43.5 0 80 23
38.5 5 130 38
33.5 10 185 54
23.5 20 305 88
13.5 30 460 133
-3.0 40.5 710 206
0.8
Note: E’=fn[(p’) , εv]; where p’ is current mean effective stress during loading and εv is vertical strain

Table 5: Stiffness non-linearity


E’/E’0.1 0.5 0.75 1.0 1.5 2.7
Strain 1.0 0.2 0.1 0.05 0.02
S0 Fill Area
Settlement with Depth

50

45 Pile head level = 43.5mOD

40

35 NSF Zone =15m

30
Elevation, mOD

Neutral Plane = 28.5mOD


25

20
Pile toe level = 16.5mOD

15

10

0
0 50 100 150 200 250
Settlement, mm

Plaxis Fill at Point B


PLAXIS Single Pile
PLAXIS Pile Group
Vdisp
NLS

Figure 8: Soil settlement prediction using NLS compared to PLAXIS at Point B

Figure 8 shows that if the thickness of soil developing NSF is assumed to be 15m (as for
Section 8.1 above), the settlement at the neutral plane would only be about 20mm. Figure 9
below compares the predicted settlements at assumed neutral plane levels for the conventional
and alternative design methods. The drag force calculated using the alternative approach is
based on an effective stress method as follows:

a. τ=kσh1’tanδ, with σho’ derived from the original Ko profile, and ∆σh’=ν’/(1- ν’)∆σv’
to allow for fill placement (with ν’=0.1), i.e. σh1’ = σho’ + ∆σh’, and k=0.8 to allow
for pile installation effects (reference CIRIA C580), or
b. τ=Kσv’tanδ, where K= (1+Ko)/2, σ’v=vertical effective stress

for both (a) and (b), δ = friction angle between pile and soil (20o).
5000 0

4000 50

Drag down, mm
Drag force, kN

3000 100

2000 150

1000 200

0 250
0 5 10 15 20
Thickness of NSF zone, m
Drag Force, Conventional Method Drag Force, Alternative Method
Drag Down, Conventional Method Drag Down, Alternative Method
Drag Down, PLAXIS

Figure 9: Comparison between conventional and alternative methods

In the absence of carrying out detailed finite element analysis, the preliminary estimates of
drag force could be applied to conventional pile group analysis methods and the resulting pile
group settlements compared. Clearly, an iterative approach is required, to achieve
displacement compatibility between the soil settlement at the neutral plane and pile group
settlement. However, the simple comparison in Figure 9 shows that a “conventional”
approach, based on a linear elastic/total stress approach would not lead to realistic predictions
of either pile group settlement or drag force.

9 Conclusions

Observations of the development of NSF on full scale pile groups are extremely rare.
However, two years of monitoring data were available for the large pile group at the eastern
arch base of the new Wembley Stadium. This enabled numerical back analysis to be carried
out to study the influence of NSF on large pile groups.

NSF on floating piles is a complex soil-structure interaction problem. However, a comparison


between two different simplified methods has given useful insights. The application of
conventional linear elastic type settlement calculations is likely to grossly overestimate the
thickness of the NSF zone and the pile group settlement (roughly equivalent to the soil
settlement at the neutral plane) would also be overestimated. Conventional total stress shaft
capacity methods would tend to overestimate the drag force.
When compared against the monitoring results and soil-structure interaction analysis, then an
alternative simplified method based on non-linear elastic settlement calculations and an
effective stress approach for calculating shaft capacity provides more plausible predictions of
pile group settlement and drag force.

The development of drag force has been shown to be very sensitive to the thickness of NSF
zone. Hence as designers, careful consideration in determining the thickness of NSF zone,
rather than the development of drag force per metre run, should be given to obtain a robust
pile design.

10 Acknowledgements

I would like to acknowledge the following people who have made this work possible:

• Imran Farooq for his initial work on the numerical modelling of the single pile
response which enabled me to progress the work to the equivalent pier simulation and
also his guidance on PLAXIS software
• David Cook for providing all the required site monitoring data from site
• Tony O’Brien for general advice and support during preparation of paper

11 References

Chandler, R. J. (2000) Clay sediments in depositional basins: the Geotechnical Cycle (The 3rd
Glossop Lecture). Quart. Jl. Engng. Geol. Hydrol., 33, pp 5-39

Chellis, R.D. (1961) Pile foundations, New York: McGraw-Hill

Farooq, I. (2002) Wembley National Stadium east arch base piling downdrag/ negative skin
friction modelling using PLAXIS, MM internal memo

Fellenius, B. H. (1998) Recent advances in the design of piles for axial loads, dragloads,
downdrag, and settlement. Proceedings of a Seminar by American Society of Civil Engineers,
ASCE and Port of New York and New Jersey, p19

Fleming, W. G. K. (1992) A new method for single pile settlement prediction and analysis,
Geotechnique, 42, No.3, pp.411-425

Fleming, W.G.K., Weltman, A.J., Randolph, M.F. & Elson, W.K. (1992) Piling Engineering
2nd Edition, E&FN Spon, London
Gaba, A.R., Simpson, B. & Powrie, W. (2003) CIRIA C580 Embedded Retaining Walls:
Guidance for Economic Design, CIRIA

Garlanger, J.E. & Lambe, T.W. (1973) Proceedings of a symposium on downdrag on piles,
Research Report R-73-56, Soil-33 Cambridge, MA: Department of Civil Engineering, MIT

Jeong, S., Lee, J.H. & Lee, C. J. (2004) Slip effect at the pile-soil interface on drag load,
Comput. Geotech. 31, No.2, pp 115-126

King, C. (1981) The Stratigraphy of the London Clay and Associated Deposits. Tertiary
Research Special Paper No.6. Rotterdam. Backhuys

Lee, C.J., Bolton, M.D. & Al-Tabbaa, A. (2002) Numerical modelling of group effects on the
distribution of dragloads in pile foundations, Geotechnique 52, No.5, pp 325-335

Lee, C.J., Lee, J.H. & Jeong, S. (2006) The influence of soil slip on negative skin friction in
pile groups connected to a cap, Geotechnique 56, No.1, pp 53-56

O’Brien, A.S., Hardy, S., Farooq, I. & Ellis, E.A. (2004) Foundation Engineering for the
UK’s New National Stadium at Wembley. Proc. 16th Int. Conf. on Soil Mechanics and
Geotech. Eng., Vol.2, p. 1533, Osaka, Japan.

O’Brien, A.S. & Sharp, P. (2001a) Settlement and heave of over consolidated clays - a
simplified non-linear method of calculation, Ground Engineering, October pp.28-32

O’Brien, A.S. & Sharp, P. (2001b) Settlement and heave of over consolidated clays - a
simplified non-linear method of calculation Part 2, Ground Engineering, November pp.48-53

Poulos, H.G. (2001) Piled raft foundations: design and applications, Geotechnique, 51, No.2,
pp.95-113

Stroud, M.A, (1989) Introduction to Standard penetration test Part 2, Proceedings of the
geotechnology conference, Institution of Civil Engineers, Birmingham July 1988, pp29-51

Vaughan, P.R. & Kwan C.Y. (1983) Weathering, structure and in-situ stress in residual soils,
Geotechnique, Vol. 34, No.1, p43-59

You might also like