Вы находитесь на странице: 1из 13

Geotechnical Engineering Proceedings of the Institution of Civil Engineers

Volume 166 Issue GE1 Geotechnical Engineering 166 February 2013 Issue GE1
Pages 18–30 http://dx.doi.org/10.1680/geng.9.00095
Validation of soil parameters for deep tube Paper 900095
tunnel assessment Received 11/12/2009 Accepted 18/11/2010
Published online 10/02/2012
Wright Keywords: geology/geotechnical engineering/tunnels & tunnelling

ICE Publishing: All rights reserved

Validation of soil parameters for


deep tube tunnel assessment
Peter J. Wright MSc, CEng, MICE, MIHT
Associate Director, Halcrow, formerly Principal Tunnel Engineer, Tube
Lines Ltd, London, UK

The paper presents the results of a desk study on London Clay and a soil investigation carried out from inside
London Underground deep tube tunnels, as part of the obligations of the Public–Private Partnership contract to
assess the deep tube tunnels. The desk study considers the geological and drainage factors affecting the coefficient
of earth pressure K0 , and its relationship with depth below ground, and some evidence from tunnel circularity data
is presented to support this. The resulting model is extended with considerations of how undrained strength varies
with depth. A site investigation conducted from inside the tunnels is described, and the results are presented and
discussed in the light of expectations from the desk study. Results for strength and stiffness testing are in general
similar to those predicted. Of particular interest are the results from piezometers installed adjacent to the tunnels.

Notation facilitate this, a cone penetration (CPT) rig was developed that
c9 effective cohesion intercept could be bolted to the cast iron tunnels to provide sufficient
E9 equivalent stiffness modulus resistance to be able to penetrate and test up to 5 m horizontally
Eu undrained stiffness modulus into the soil. In addition, cone pressuremeter tests were per-
G shear modulus formed, samples were obtained for laboratory testing and piezo-
Gmax maximum shear modulus at very low strain meters were installed, generally at 1, 3 and 5 m horizontally from
K9 ratio of total horizontal stress on tunnel axis to total the tunnel extrados at a number of locations.
vertical stress on tunnel crown
K0 coefficient of earth pressure at rest (effective stresses) 2. Desk study
Knc coefficient of earth pressure for normal consolidation In order to carry out the most basic analysis of a tunnel using the
p9 (ó v9 þ 2ó h9 )=3 elastic continuum method, the soil parameters typically required
Su undrained strength in triaxial compression are the in situ stresses (and the coefficient of earth pressure, K0 )
z depth below top of London Clay and the soil stiffness (e.g. Young’s modulus, which is frequently
í9 effective Poisson’s ratio related directly to the soil strength).
ó h9 horizontal effective stress
ó v9 vertical effective stress 2.1 Coefficient of earth pressure at rest, K0
ö9 effective angle of friction The coefficient of earth pressure at rest, K0 , is defined as the
horizontal to vertical effective stress ratio. K0 depends on the
1. Introduction stresses that the material has been subjected to as a result of
Under the Public–Private Partnership (PPP) contract with London the geological and pore water pressure history. Using a spread-
Underground, Tube Lines was required to improve the knowledge sheet model, the results for K0 at various stages in the geological
and understanding of the deep tube tunnel assets of the Jubilee, history of London Clay have been calculated, based on the
Northern and Piccadilly Lines. These are the tunnels that have following methodology.
been constructed by mining methods, as opposed to cut-and-
cover, and which vary in depth below ground from approximately First, London Clay was deposited in a shallow sea environment in
5 to 80 m. Clearly, it was important to understand any changes in the early Eocene age. The full sequence of London Clay was in
the surrounding London Clay and other subsoils that had oc- the order of 90 m thick (at Reading) to 150 m thick (in Essex)
curred since construction, up to 120 years ago. In particular, in according to King (1981). In central London the thickness has
order to carry out an assessment of the tunnels, an understanding been taken to be about 120 m, as shown in Figure 1.
of current in situ soil stresses, pore water pressures, strength and
stiffness was required. Further deposition of other strata over the London Clay then
followed. The formula proposed by Jaky (1944) can be used to
For this reason, a desk study was carried out, and subsequently estimate the normally consolidated earth pressure coefficient Knc
further investigation of the soil around the tunnel was undertaken for this original deposition of the London Clay, using the expres-
through holes diamond-drilled through the tunnel linings. To sion

18
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

Grain size
deposits and a varying proportion of the London Clay (depending
on location) were eroded. A remnant of the Bagshot Beds was
100 m
left behind at Hampstead, on top of the London Clay, which
Sandy clay
implies that the full thickness of London Clay can be found on
E Hampstead Heath, but as no records of boreholes penetrating to
90 m the base of the London Clay have been found at this location,
Silty clay there is no direct measurement of thickness of the strata to verify
this. The full depth of erosion of London Clay and its overlying
Silty clay with strata in central London has been estimated to be between 152
D2 sand partings
80 m and 396 m, according to a review by Pantelidou and Simpson
G G G G Silty clay
(2007). An analysis of K0 during this erosion process can be
Sandy clay modelled, based on information from Burland et al. (1979) and
D1b
Silty clay Simpson et al. (1989), to provide a profile of K0 with depth below
70 m the top of the London Clay strata.
G
G
G Sandy glauconitic
G clay
D1a First, the changes in effective stress due to erosion can be
Silty clay modelled. Assuming the water table is at the surface throughout
60 m the erosion process, the removal of 200 m of overburden would
Silty clay with
sand partings effectively result in a reduction in vertical effective stress of
C2 approximately 2000 kPa. Horizontal effective stresses tend to
Silty clay remain locked in, but reduce to some extent as a result of the
50 m G G G G
Poisson’s ratio effect of this unloading. Mayne and Kulhawy
Homogeneous
C1 fissured clay (1982) proposed that the change in K0 over this time can be
G G G G related to the overconsolidation ratio (OCR) according to
Clay with
40 m silt/sand partings
K0
B2 ¼ OCR m
Homogeneous 2: K nc
fissured clay
30 m

Homogeneous where m ¼ sin ö9.


B1 fissured clay
G G G G
20 m According to Mohr–Coulomb theory, passive plastic yielding due
Clay with to untenable horizontal stresses near the surface is governed by
silt/sand partings
A3 the equation
Homogeneous
fissured clay
10 m   sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ sin ö9 1 þ sin ö9
Interbedded ó h9 ¼ ó v9 þ 2c9
A2 silty clay and 3: 1  sin ö9 1  sin ö9
sandy clay

0
G: glauconite
These relationships have been plotted in Figure 2 for the removal
of 200 m of overburden, assuming c9 ¼ 5 kPa.
Figure 1. King’s stratigraphic sequence for London Clay (courtesy
of Chris King)
The deposition of drift deposits (mainly Thames Gravels) of up
to 15 m depth followed in many locations. To model the
deposition of drift deposits on top of the London Clay, the effect
is mainly an elastic one (Burland et al., 1979), so the Poisson’s
1: K nc ¼ 1  sin ö9
ratio effect on horizontal stresses would be

Assuming a London Clay with a drained friction angle ö9 ¼ 238, ˜ó h9 í9


¼
a value of Knc is obtained that is initially constant and equal to 4: ˜ó v9 1  í9
0.61. This is plotted as a line parallel to the y-axis in Figure 2.

During the late Tertiary and Pleistocene eras, the overlying Simpson et al. (1989) has suggested that a sensible value for this

19
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

K0
0 1 2 3 4 5
0

Jaky's eqn
Depth below top of London Clay: m

10

K0 (unload)
20
K0 (reload)
30
K0 (underdrained)

40
K0 (50% underdrain)

50 K0 (No. 1 Poultry)

60

Figure 2. K0 Variation with depth

result would be 0.33 (implying í9 ¼ 0.25), and that is the value programme it was not understood in what situations tunnels could
that has been used in the model, and plotted in Figure 2. be considered permeable or impermeable, as a site investigation at
the Kennington loop tunnel, reported by Gourvenec et al. (2005),
More recently, over the last 100 years or so, London Clay has been showed that that particular tunnel did not appear to be permeable,
partially underdrained as a result of extraction from deep wells. as near-hydrostatic pressures were measured in a borehole very
The subsequent reduction in pore water pressure has resulted in close to the tunnel. These measurements were, however, at the
further increases in the vertical effective stress and, to a much base of the London Clay, in more permeable material.
lesser extent, in the horizontal effective stress. The consequent
further drop in K0 is also shown in Figure 2, as is the curve for Drainage of water from the London Clay into a tunnel tends to
50% underdrainage (reduction to half hydrostatic pore pressures). increase the vertical effective stress on the tunnel crown when
This latter curve corresponds quite well to a K0 profile measured compared with the axis, causing an ovalisation, or ‘squat’, of the
at Bank, part of the site investigation at No. 1 Poultry. However, in tunnel. This is nevertheless complex, and a tunnel ovalisation
reality the degree of underdrainage would be more pronounced may at least in part be as a result of the construction conditions.
near the drainage source (e.g. the bottom of the London Clay), and
in fact the measurements at No. 1 Poultry do tend to coincide with A significant number of circularity surveys have been carried out
the 100% underdrainage curve at the deepest depth in Figure 2. in the running tunnels of the Jubilee, Northern and Piccadilly
On this basis, the model seems a reasonably good one. Lines. Almost invariably these surveys have shown that there is
indeed a shortening of the vertical axis of the tunnels, that is, a
2.2 Pore water drainage effects on tunnel circularity tunnel ‘squat’. The exceptions tend to be where there have been
Most site investigations adjacent to tube tunnels in London show construction difficulties, or where the tunnel permeability has
some degree of reduction from a hydrostatic pore water pressure been reduced by grouting, for instance near Kennington, and
profile due to underdrainage. Figure 3(a) shows a typical pressure where tunnels cross under the River Thames. This is clearly
distribution from a site investigation at Waterloo station. While a shown in Figure 4, which shows a longitudinal plot of distortion
significant proportion of this underdrainage is from the base of the of the tunnel vertical axis from Embankment to Waterloo
London Clay (as a result of historic lowering of groundwater due stations, where the tunnels tend to be more circular under the
to abstraction, which continues to be maintained by Thames Water river, and ‘squat’ elsewhere. This is considered to be as a result
under its Gardit programme), some of the underdrainage is thought of the tunnels under the river being periodically grouted and
to be due to the permeability of the tunnels themselves, draining therefore comparatively watertight, compared with the remainder
small quantities of water from the London Clay. This is illustrated of the tunnels, which are thought to act as drains.
by Figure 3(b), from the interpretive report for a site investigation
adjacent to Camden Town station, which shows a pore water Another incidental point of note is that one of two twin tunnels
pressure close to zero at the level of the tube tunnels themselves, usually tends to ‘squat’ more than the other, and this is assumed
and the profile of pore water pressure with depth has been inferred to be the first tunnel constructed, which subsequently squats more
as shown. However, at the start of the tunnel assessment owing to relief of horizontal ground stresses as the adjacent

20
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

Initial pore water pressure: kPa


that the northbound road, showing greater distortion, would have
0 50 100 150 200 250 300 350 400
10 been the first tunnel constructed.

Although there are instances where the tunnel has been built more
0
out of shape, in general average tunnel squat appears to be
between 0.5% and 1% of diameter across all tube tunnels
⫺10 surveyed. While there is no proof that the tunnels were not built
like this, it seems sensible to assume that the tunnellers of the day
⫺20 would have attempted to build a circular lining out of the back of
Level: m

Hydrostatic
the tunnelling shield (and this circular ring would have been
achieved on average), and that the squat is subsequently caused
⫺30
by stresses coming onto the lining after construction as the result
of a combination of short-term and long-term soil pressures.
⫺40
For the combined short- and long-term case, therefore, it seems
⫺50 reasonable to assume that the ratio of horizontal stress acting at
the tunnel axis to vertical stress acting on the crown of a tunnel is
generally less than unity. In fact Tube Lines’ basic assessment of
⫺60 (a) deep tube tunnels using the elastic continuum method (for the
Water pressure: kPa long term) assumes that K9 (total stresses) is 0.7. However,
0 50 100 150 200 250 300 350 assessments using finite-element methods have assumed that K0
30
Piezometers is higher, at 1.0 to 1.25 prior to tunnel construction, and that the
25 Water strike pore water pressure changes due to drainage into a permeable
Morning water level tunnel in the long term that are responsible for the changes in
20 Design line effective stress causing tunnel distortion, are explicitly modelled.

15 2.3 Soil undrained strength Su


Clearly, the undrained strength of London Clay can be related to
10 the degree of consolidation, and the model can easily be
extended, based on Mohr’s circle considerations, to give the
5 relationship between the undrained compressive strength with the
in situ effective stresses, and hence the depth below the ground
Elevation: mOD

0 surface. Assuming that the horizontal effective stress remains


constant, the measured undrained strength Su for consolidated
⫺5 undrained compression can be determined at the point when the
vertical effective stress has increased so that the Mohr’s circle
⫺10 touches the failure envelope. It can easily be shown from Mohr’s
Approx. elevation
of tube tunnels circle geometry that
⫺15
 
ó h9 sin ö9 þ c9 cos ö9
⫺20 Su ¼ ôðfailureÞ ¼ cos ö9
5: 1  sin ö9
⫺25

The resulting profiles for Su are shown in Figure 6. A profile


⫺30
typically used for Su in London Clay of 50 + 8z kPa is also shown
⫺35 for comparison. It is worth noting that Su increases quite rapidly
(b) with depth near the surface of the clay, as a result of the high
Figure 3. Pore water pressure variation with depth and LU overconsolidation ratio and high K0 values, but that the rate slows
stations: (a) Waterloo; (b) Camden Town to about 5 kPa per metre depth below about 15 m, according to
this model. Effects of full underdrainage are also shown; as
expected, this leads to a further strengthening of the clay. This
tunnel is constructed. This is illustrated more clearly in Figure 5, graph is not intended to give any design or assessment para-
a similar longitudinal plot between Golders Green portal and meters, as other factors, such as fissuring and variations in
Hampstead, where the tunnels have been constructed closer than structure, clearly have a significant effect, but this does illustrate
those between Embankment and Waterloo, and where it appears general trends. Of particular note is the effect of drainage, which

21
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

Southbound ring number


1200
1160
1120
1080
1040
1000
960
920
880
840
800
760
720
680
640
600
560
520
480
440
400
360
320
280
240
200
160
120
80
40
0
2 2

Approx. extent of river

1 1
% ooc (squat ⫹ve)

0 0

Northbound
⫺1 Southbound ⫺1
S/B moving average
Waterloo station N/B moving average Embankment station

⫺2 ⫺2
0
40
80
120
160
200
240
280
320
360
400
440
480
520
560
600
640
680
720
760
800
840
880
920
960
1000
1040
1080
1120
1160
1200
Northbound ring number

Figure 4. Plot of tunnel distortions between Embankment and


Waterloo

Southbound ring number


2560
2520
2480
2440
2400
2360
2320
2280
2240
2200
2160
2120
2080
2040
2000
1960
1920
1880
1840
1800
1760
1720
1680
1640
1600
1560
1520
1480
1440
1400
1360
1320
1280
2 2
% ooc (squat ⫹ve)

1 1

0 0

⫺1 ⫺1
Northbound
Southbound ⫺2
1100
1140
1180
1220
1260
1300
1340
1380
1420
1460
1500
1540
1580
1620
1660
1700
1740
1780
1820
1860
1900
1940
1980
2020
2060
2100
2140

S/B moving average


N/B moving average Northbound ring number

Figure 5. Plot of tunnel distortions between Golders Green and


Hampstead

appears to have a significant effect on undrained strength, London Clay around tunnels. For the initial assessment of
particularly at depth. tunnels, Tube Lines has taken the undrained stiffness for London
Clay used in numerical modelling, Eu , to be 400Su :
2.4 Soil stiffness
Soil stiffness is another difficult parameter to assess, as it is For a tunnel construction 20 m deep, with K0 ¼ 1 and the water
known to vary according to the amount of strain to which it is table at ground surface, p9 would be 200 kPa. Assuming Su ¼
subjected, according to Jardine et al. (1986). Typical profiles 50 + 8z ¼ 210 kPa, where z is the depth below the London Clay
showing the variation of shear stiffness G (assumed to equal E/3 surface, Eu would be 400 3 210 ¼ 84 000 kPa, and Eu /p9 ¼ 420,
for undrained conditions) with strain in London Clay are shown implying strains in the range shown in Figure 7.
in Figure 7, from Hight and Higgins (1994).
While it is understood that a volume of soil within a diameter or
A review of data used on the Jubilee Line Extension Project, and so of the tunnels will have been strained during the construction
used by consultants on outside party schemes, suggests that an process, to the extent that the original stiffness is reduced as in
undrained Young’s modulus (Eu ) of 250 to 600 times Su is a Figure 7, it was not known whether, or to what value, the stiffness
reasonable assumption for predictions of ground deformations in of the soil might have recovered during the period since construc-

22
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

Su: kN/m2
0 200 400 600 800 1000
0

10
Depth below top of London Clay: m

Deposition
20
Erosion of 200 m
30
Drift deposition
of 10 m
40
50 ⫹ 8z
50 Underdrainage
(full hydrostatic)
60

70

80

Figure 6. Undrained strength modelling during geological history

2500 needed to be returned to the state of stress they were subject to


Range of measurements prior to sampling. One way to estimate this was to carry out a
(Hight and Higgins, 1994)
2000 finite-element analysis of the tunnel construction and subsequent
consolidation effects over the time since construction, prior to
Assumed strain range
1500 testing. This is described later in this paper in Section 3.4.
for tunnel construction
3G/p⬘

1000
3.1 Preparation for laboratory testing
Sampling was carried out by driving in a 72 mm diameter Mostap
sampler, and for two of the sites a 150 mm diameter thin-walled
500
sampling tube was used. Sample quality was evaluated by using
X-ray tomography. Test samples for triaxial testing were then
0
0·00001 0·0001 0·001 0·01 0·1 1 carefully cut more or less perpendicularly from the subhorizontal
Axial strain, ea: % sample in a high-humidity environment, so that triaxial samples,
51 mm in diameter by 51 mm high, were obtained, oriented
Figure 7. Variation of London Clay shear stiffness in compression vertically. While an aspect ratio of 1.0 may be unusual in UK
with strain practice, it is generally regarded as acceptable, when the sample
ends are lubricated, for simulation of pre-peak behaviour.

tion of the tunnels, during which consolidation of this volume of 3.2 Soil classification tests
soil would have been expected to have caused an increase in Testing indicates that the bulk unit weight of the London Clay
stiffness. One of the main purposes of the site investigation was approximates to 20 kN/m3 (Figure 9). As might be expected,
to determine this. there is a general trend of increasing unit weight with depth, as a
result of consolidation during geological history, but the deepest
3. Site investigation samples taken from Golders Green 1 below Hampstead Heath
A site investigation was carried out from inside the existing have a slightly lower density than might be expected.
running tunnels. Table 1 shows the locations of cone penetration
tests (CPT), cone pressuremeter tests (DPM), sampling (Lab) and Figure 10 shows the sand/silt/clay distributions, water contents
piezometer installation (Piez), and Figure 8 shows the CPT and Atterberg limits, and the only point worthy of note is that the
apparatus fixed to a tunnel lining. samples from the Lambeth Group strata at Bond Street have a
significantly higher proportion of clay, but this does not seem to
One of the initial difficulties of this work is that it was unlike a correspond to a particularly high plasticity index.
normal site investigation, in that the holes made were horizontal
or subhorizontal at up to 198. In addition, the state of stress in the 3.3 Piezometers
soil was complicated by the stresses induced by the existence of Three piezometers (both vibrating wire and BAT pressure types
the tunnels. For instance, for the laboratory testing the samples were used) were installed at distances of 1, 3 and 5 m horizon-

23
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

Station of entry Date Tests Geology Depth: m

From From top of


surface LC strata

Aldwych trial May 2004 CPT, DPM London Clay (A2/A3) 30.7 25.2
Tooting Bec 10 Oct 2004 CPT, DPM, Lab London Clay (A2) 16.6 11.8
Arsenal 19 Oct 2004 CPT, DPM, Lab Piez London Clay (A3) 9.0 4.0
Oval 26 Oct 2004 CPT, DPM, Lab Piez London Clay (A3) 14.9 7.1
Bond Street 3 Nov 2004 CPT, DPM, Lab Piez Lambeth Group 31.0 26.5
Golders Green I 9 Nov 2004 CPT, DPM, Lab Piez London Clay (C3/D1) 38.6 38.6
Golders Green II 15 Nov 2004 CPT, DPM, Piez London Clay (C3/D1) 64.8 64.8
Covent Garden 30 Nov 2004 CPT, DPM London Clay (A2) 35.6 28.8
Aldwych 19 Jan 2005 CPT, DPM, Lab Piez, Seismic London Clay (A3) 27.6 20.9
Angel Mar 2005 CPT, DPM Lambeth Group 36.5 36.5
Swiss Cottage 1 5 Sep 2005 CPT, DPM, Piez London Clay (Unit?) 7.1 7.1
Swiss Cottage 2 5 Sep 2005 CPT, DPM, Piez London Clay (Unit?) 10.1 10.1
Green Park 1 11 Sep 2005 CPT, DPM, Piez London Clay (B1) 23.5 12.2
Green Park 2 11 Sep 2005 CPT, DPM, Piez London Clay (B1) 27.0 14.9
Warren St 1 18 Sep 2005 CPT, DPM, Piez Lambeth Group 23.5 12.3
Warren St 2 18 sep 2005 CPT, DPM, Piez Lambeth Group 29.1 16.4
Leicester Sq 1 25 Sep 2005 CPT, DPM, Piez London Clay (A3/B1) 21.1 18.4
Leicester Sq 2 25 Sep 2005 CPT, DPM, Piez London Clay (A3) 25.1 19.9
Heathrow 1 3 Oct 2005 CPT, DPM, Piez London Clay (Unit?) 12 12
Heathrow 2 3 Oct 2005 CPT, DPM, Piez London Clay (Unit?) 10 10

Table 1. Testing locations

Unit weight: kN/m3


19 19·5 20 20·5 21
0

5
Depth below ground: m

10 Arsenal

15 Oval
Tooting
20

25 Aldwych
30
Bond St
35
Golders 1
40
Figure 8. CPT apparatus in tunnel
Figure 9. Soil unit weight

tally from the tunnel at 10 locations around the Tube Lines


network. After installation they took a period of several months the inflow of water is limited, and the piezometers have to be
to reach equilibrium values after dissipation of excess pore water perfectly sealed in order to measure the gradient of pore water
pressures resulting from the installation process. From Figure 11 pressure accurately. It may be that the piezometers showing low
it can be seen that, in general, pore pressures are low or declining pressures were not perfectly sealed, but equally the shallower
towards the tunnel, indicating that groundwater is draining into pore pressure drawdown may be due to the existence of more
the tunnel. Needless to say, as London Clay is very impermeable, permeable lenses in the clay.

24
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

Composition: %
results, and tended to overestimate the in situ pressure. In
0 20 40 60 80 100
0 particular, the DPM estimated in situ stresses of up to three times
5 the overburden pressure. It is thought that the excess pore water
Depth below ground: m

10
pressures set up by pushing the cone in do not have time to
Sand dissipate in the limited time available for measurement, and this
15
Clay is the cause of the higher-than-expected measurements.
20
Silt
25
In order to model in situ stresses for the consolidation phase of
30 Bond St the consolidated undrained tests to determine Su , the assumption
35 was made that K0 varied from 0.5 at the tunnel axis to 0.9 at a
40 position 5 m horizontally from the tunnel axis. Fully coupled
(a)
finite-element analyses of tunnel construction and 100 years in
WC and Atterberg limits: % use were also carried out to verify these assumptions. In all cases
0 20 40 60 80 100 the tunnel was assumed to act as a drain. Table 2 shows some in
0
situ stresses from the finite element analysis for the site at the
Depth below ground: m

5 Golders Green 1 site under Hampstead Heath, together with the


10 assumed K0 values used in the testing. The assumed values
15 WC correspond reasonably well with the calculated values.
20 LL
25 PL
3.5 Undrained shear strength Su
30 PI
Various techniques were used to estimate Su from the site
35 investigation. From the subhorizontal samples extracted from the
40 six sites, the vertical samples obtained were tested in consolidated
(b) undrained triaxial compression. In addition, CPT tests, driven
Figure 10. Soil classification: (a) particle size; (b) water content horizontally from the tunnel, were correlated with undrained
and Atterberg limits (WC ¼ water content; LL ¼ liquid limit; strength, and CPT pressuremeter tests (DPM) were also corre-
PL ¼ plastic limit; PI ¼ plasticity index) lated with Su using the Houlsby and Withers (1988) method. It
should be noted that while the laboratory tests gave a result for
the undrained strength in the vertical direction (with the excep-
tion of one of the Aldwych samples), CPT measurements should
3.4 In situ stresses correlate with the horizontal undrained strength, and pressure-
In situ stresses were determined in the laboratory from soil meter tests should correlate with the average of the vertical and
suction measurements, and in situ using the cone pressuremeter out-of-plane horizontal strength. Figure 12 summarises the result-
(DPM). Both methods showed a high degree of scatter in the ing measured average undrained strengths from all three methods.

200
Aldwych

Golders Green 2

150 Golders Green 1


Pore water pressure: kPa

Oval

Green Park
100
Leicester Square

Bond Street

50 Heathrow T4

Goodge Street

Swiss Cottage
0
1 3 5
Horizontal distance from tunnel wall: m

Figure 11. Stabilised piezometer readings

25
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

Horizontal FE pwp Measured pwp FE vertical FE horizontal FE Resulting K0 from Assumed K0 for
distance from above above tunnel effective stress: effective stress: out-of-plane FE, ignoring consolidation
tunnel axis: m tunnel axis: m kPa (%OB) kPa effective out-of-plane phase of triaxial
axis: kPa stress: kPa stress tests

1 45 40 750 (98) 473 477 0.63 0.58


3 110 130 690 (90) 497 470 0.72 0.74
5 155 160 630 (82) 489 457 0.78 0.9

Table 2. Finite-element (FE) analysis and measured parameters


for Golders Green 1

5
Depth below top of London Clay: m

10 CPT
DPM
15
Lab
20 Su(DPM) ⫽ 113 ⫹ 12·0z
Su(CPT) ⫽ 57 ⫹ 10·3z 50 ⫹ 8z
25
Linear (CPT)
30 Linear (DPM)
Su(Lab) ⫽ 57 ⫹ 12·9z
35 Linear (Lab)

40

45
0 100 200 300 400 500 600
Su: kPa

Figure 12. Shear strength against depth from in situ and


laboratory testing, excluding Golders Green samples

It might be expected that the horizontal CPT tests would result in Up to 36 m deep
lower measured strengths, owing to the estimated in situ stress
Golders
regime with K0 approximating 0.7 at the test position, and this is
50 ⫹ 8z
indeed the case. The scatter across both in situ tests is relatively
Depth below top of London Clay: m

high, but the four laboratory tests shown lie closer to a straight 0
line. If all the results (including the tests between Golders Green 10
and Hampstead) are combined on a single graph (Figure 13), the
20
correlation of Su with depth (neglecting CPT and pressuremeter
results at the deepest sites near Hampstead) can be compared 30
with the Su ¼ 50 + 8z profile assumed, which seems to represent a 40
conservative lower bound to the results. However, the data from 50
the Golders 1 and 2 sites do undoubtedly show a lower undrained
60
strength than would be predicted from the trend of results at
70
lower depths, and this is considered further below. 0 100 200 300 400 500 600
Su: kPa
3.6 Stiffness
Pressuremeter tests were carried out at each of the running tunnel Figure 13. All tests for Su in London Clay
locations, using unload–reload cycles to obtain a range of shear

26
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

modulus (G) values. In addition, laboratory tests using bender The normalised values for 3Gmax /p9 correspond to about 2400,
elements during the consolidation stage of the triaxial tests were which is right at the upper end of the stiffnesses from Figure 7. This
used to determine Gmax , which is the highest value of measured suggests that any reductions in stiffness resulting from the strains
shear modulus at very small strains. Figure 14 summarises these of the original construction process have been completely reset.
data.
From the data analysed, there is no reliable trend to suggest that
Tests for shear modulus Gmax in the laboratory and for G using the ground adjacent to the tunnel wall has a significantly different
the pressuremeter on site are shown in Figure 14. Also shown is a stiffness than the ground further away from the tunnel.
plot of G (assumed to equal E/3), where E ¼ 0.4(50 + 8z) MPa,
the preliminary value initially assumed for the tunnel assess- 3.7 Permeability
ments, which correlates well with the measured DPM values. Permeability was measured twice during the consolidation stage of
However, in a similar manner to values for undrained strength, the triaxial test, and again in the oedometer test. Average values
the two Golders Green sites are not well represented by this are plotted in Figure 15, which shows permeability decreasing
formula. with depth, as would be expected, with the exception of the sample

10
Depth below top of London Clay: m

20 Gmax(Lab) ⫽ 24 ⫹ 5·7z G (DPM)


Gmax (Lab)
30
G (DPM) ⫽ 0·9 ⫹ 1·7z Golders (DPM)
40 Golders (Lab)
50 0·4(50 ⫹ 8z)/3
Linear (G (DPM))
60
Linear (Gmax (Lab))
70 Linear (Golders (DPM))
80

90
0 50 100 150
G: MPa

Figure 14. Undrained stiffness from pressuremeter and


laboratory testing

5
Depth below ground surface: m

10

15

20

25
Golders Green 1
30

35

40

45
0 2·00 ⫻ 10⫺12 4·00 ⫻ 10⫺12 6·00 ⫻ 10⫺12 8·00 ⫻ 10⫺12 1·20 ⫻ 10⫺11 1·400 ⫻ 10⫺11 1·00 ⫻ 10⫺11

Permeability: m/s

Figure 15. Average permeability at each site measured from


laboratory tests

27
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

Gvh: MPa
from under Hampstead Heath (Golders Green 1), which is
0 40 80 120 160 200 240 280
significantly more permeable, although values are all very low, 0
ranging from 4 3 1012 to 1.2 3 1011 : These values are at the Brent Cross
Canon’s Park
lower end of the range normally assumed for London Clay. T5
Finchley
4. Commentary on results with respect to Blackwall
10 Vauxhall
assessment Moor House

Depth below ground level: m


While the overall results are encouraging, in that the initial Chattenden
trendline assumptions about strength and stiffness seem to be
justified, measured strength at the two sites under Hampstead
Heath (Golders 1 and Golders 2) are significantly below the 20
trendline Su ¼ 50 + 8z kPa, and stiffness is correspondingly re- Unit B
duced. In addition, the lower density and higher permeability at
Golders Green seem to suggest a more open grading than at the
other sites. The assessments at these locations need to be Unit A3
revisited, and other locations that would be expected to display 30
the same differences should be identified.

Hight (2004) has identified differences in stiffness in the various


London Clay strata by analysing data from various sites. It has
40
been stated by King and others, as shown in Figure 1, that each
layer becomes coarser with elevation, and this can be correlated
with changes in water content. Figure 16 illustrates this, and from Figure 17. Summary of Gvh against depth for different London
Figure 17 it is evident that unit A3 (20–30 m from the base of Clay units
the London Clay), for instance, does seem to have a lower

Water content: % Water content: % Water content: %


16 20 24 28 32 16 20 24 28 32 16 20 24 28 32
0 0 0

B B B

20
Match by unit boundary: m

Match by unit boundary: m

20 20

A3
Depth: m

A3 A3

A2
A2 A2
40 40 40

LG LG

60 60 60
St James’s Park St James’s Park
Waterloo International Terminal Bishopsbridge
Waterloo JLE

Figure 16. Trends in water content between London Clay units

28
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

stiffness gradient with depth and a lower water content than do occurred at Golders Green 2 and 120 m at Golders Green 1,
the other units. This is not so evident in greyscale, but a rough compared with 200 m elsewhere. If the original deposition of
trendline has been drawn on Hight’s figure to illustrate this. It is London Clay is assumed to be horizontal, then Su can be related
likely therefore that different units have different strength and to ground surface elevation as well as tunnel depth (Figure 18).
stiffness profiles. This is a purely illustrative spreadsheet model, and is not
calibrated to the actual measurements of undrained strength, but
As previously stated, no boreholes have been identified that go to on this basis one might expect to obtain a 15–20% decrease in
the bottom of the London Clay at Hampstead Heath, so that level undrained strength for a tunnel at the same depth but where the
is unknown. Existing boreholes suggest the top of the London ground level is 100 m higher, or 0.2% decrease per metre
Clay is between 103 and 107 m AOD at these locations. elevation. So, in summary, the reduced strength is due to a lower
Additionally the spring line, evident from OS maps and assumed overconsolidation ratio (OCR), owing to less erosion of overlying
to mark the top of the London Clay, is at about 105 m AOD. strata occurring at Hampstead Heath.
Golders 1 and Golders 2, at levels of 70 m and 65 m AOD,
appear to be quite high in the sequence, in unit C3 or D1. Using this model, Su would be expected to be about 250 kPa at
However, there are little data available for these units, as they are Golders 1 and 360 kPa at Golders 2, which is quite a good fit to
mostly eroded in central London. However, this layer of London the test data. However, it should be noted that, on the basis of the
Clay may well have a lower strength and stiffness, associated with drainage profile around the tunnel from Figure 11, it might be
being slightly less dense, coarser and more permeable than would expected that the strength adjacent to the tunnel would be
be expected at this depth below ground. increased owing to the increase in vertical effective stress after
pore water drainage into the tunnel has equilibrated, as illustrated
Another possible partial explanation for the reduced strength at in Figure 6.
this location is illustrated by the principle in Figure 6. The
trendline Su ¼ 50 + 8z kPa may be reasonable for the first 15 m 5. Conclusion
below the surface of the clay, but below that the degree of Tube Lines, as part of its tunnel assessment work, has had to
overconsolidation due to erosion is reduced, and the gradient of make and justify assumptions about the generic strength and
increasing strength becomes less. stiffness of the London Clay strata in central London up to 120
years after the tube tunnels were constructed.
A further partial explanation of the lower strength at these two
sites is related to the amount of erosion that has occurred on As part of a desk study, a spreadsheet model has been used to
Hampstead Heath, where, on top of the Heath, geological erosion understand trends in K0 and Su with depth below ground due to
could be around 100 m, or 50% less than at the other sites. If, in deposition, erosion, redeposition and underdrainage. This spread-
the spreadsheet model, we now take the ground surface to be sheet model can be used to predict in situ stresses and strengths
100 m higher, and reduce the geological erosion to 100 m, Figure of the London Clay, based on geological erosion, deposition and
6 will be redrawn; the original strength after deposition will be drainage conditions.
lower, but so also will be the reduction in strength after erosion.
The net effect is a reduction in strength relative to depth. This A site investigation has been carried out from within the tunnels,
can be illustrated, assuming that only 100 m of erosion has and useful information has been obtained on groundwater drain-

Low elevations
e.g. central London Golders 1 Golders 2 Pre-erosion

500
450
400
10 m deep
350
300 30 m deep
Su: kPa

250
50 m deep
200
150 70 m deep
100
50
0
0 50 100 150 200 250
Ground elevation: mOD

Figure 18. Variation of Su with different geological erosion

29
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering Validation of soil parameters for deep
Volume 166 Issue GE1 tube tunnel assessment
Wright

age adjacent to the tunnels, and on strength, stiffness and of the Institution of Civil Engineers – Geotechnical
permeability, and their variation with depth below ground. Engineering 158(1): 25–33.
Hight DW (2004) Characterisation of London Clay. Unpublished
While, in general, the desk study and site investigation justify the Technical Report to Tube Lines Ltd, Geotechnical Consulting
use of the generic parameters assumed for the assessment of Group, London, UK (Tube Lines document reference: GCG-
tunnels in London Clay, the two testing sites under Hampstead L001-N416-DTAAWP2-TUN-RPT-00001).
Heath appear to have a lower soil density, undrained strength and Hight DW and Higgins KG (1994) An approach to the prediction
stiffness than might be expected, and a higher permeability. To of ground movements in engineering practice: background
some extent the spreadsheet model could predict these lower and application. Keynote Lecture. Proceedings of the
strengths, but it is likely that the structural nature of the subsoil International Symposium on Pre-failure Deformation
(apparently a slightly sandier clay in unit C3 or D1 of the London Characteristics of Geomaterials, Hokkaido, Japan, pp. 909–
Clay) contributes to the lower strength and stiffness. 945.
Houlsby GT and Withers NJ (1988) Analysis of the cone
As a result of the desk study and site investigation, the pressuremeter test in clay. Géotechnique 38(4): 575–587.
parameters used in Tube Lines’ assessments have been justified, Jaky J (1944) The coefficient of earth pressure at rest. Journal of
but assessment of the deeper tunnels in London Clay may need to the Society of Hungarian Architects and Engineers 78(22):
be revisited, to determine whether a more conservative soil 355–358.
strength and stiffness need to be adopted for the assessment. Jardine RJ, Potts DM, Fourie AB and Burland JB (1986) Studies
of the influence of non-linear stress–strain characteristics in
Acknowledgements soil–structure interaction. Géotechnique 36(3): 377–396.
The author would like to thank Lankelma (particularly Andy King C (1981) The Stratigraphy of the London Clay and
Barwise and Darren Ward) for developing and operating the CPT Associated Deposits. Backhuys, Rotterdam, the Netherlands,
kit, SeTech for carrying out the DPM tests, CMCS (Gerwyn Price), Tertiary Research Special Paper No. 6.
for installation of piezometers and GCG (David Hight and Kelvin Mayne PW and Kulhawy FH (1982) K0 –OCR relationships in
Higgins) for a number of supporting studies and advice. Finally, soil. Journal of Geotechnical Engineering, ASCE 108(GT6):
thanks to my mentor Ivan Chudleigh for stimulating my interest. 851–872.
Pantelidou H and Simpson B (2007) Geotechnical variation of
REFERENCES London Clay across central London. Géotechnique 57(1):
Burland JB, Simpson B and St John HD (1979) Movement 101–112.
around excavations in London Clay. Proceedings of the 7th Simpson B, Blower T, Craig RN and Wilkinson WB (1989) The
European Conference on Soil Mechanics, Brighton, UK, pp. Engineering Implications of Rising Groundwater Levels in
13–29. the Deep Aquifer Beneath London. Construction Industry
Gourvenec SM, Mair RJ, Bolton MD and Soga K (2005) Ground Research and Information Association, London, UK, CIRIA
conditions around an old tunnel in London Clay. Proceedings Special Report 69.

WHAT DO YOU THINK?


To discuss this paper, please email up to 500 words to the
editor at journals@ice.org.uk. Your contribution will be
forwarded to the author(s) for a reply and, if considered
appropriate by the editorial panel, will be published as a
discussion in a future issue of the journal.
Proceedings journals rely entirely on contributions sent in
by civil engineering professionals, academics and students.
Papers should be 2000–5000 words long (briefing papers
should be 1000–2000 words long), with adequate illustra-
tions and references. You can submit your paper online via
www.icevirtuallibrary.com/content/journals, where you
will also find detailed author guidelines.

30
Downloaded by [ UNIVERSITY OF IOWA LIBRARIES] on [11/09/16]. Copyright © ICE Publishing, all rights reserved.

Вам также может понравиться