Professional Documents
Culture Documents
Concerete Retaining Wall
Concerete Retaining Wall
Civil Engineers
Geotechnical Engineering 163
October 2010 Issue GE5
Pages 279–290
doi: 10.1680/geng.2010.163.5.279 Howard Roscoe David Twine
Principal Geotechnical Director, Ove Arup and
Paper GE-D-09-00052 Engineer, Atkins Design Partners, London, UK
Received 26/06/2009 Environment & Engineering,
Accepted 06/07/2009 Risley, UK
Keywords: design methods & aids/
field testing & monitoring/retaining
walls
Embedded retaining walls for 1.8 km of cut-and-cover developed for use throughout the Channel Tunnel Rail Link
tunnels and earth-retaining structures were redesigned (CTRL) project, now known as High Speed 1 and much of this
during the construction period to optimise construction is incorporated in recent guidance on the design of embedded
methods and temporary propping. The design approach walls (Ciria report C580: see Gaba et al., 2003).
included many of the developments now recommended
in Ciria report C580, and the paper summarises the This paper describes the geometry of 14 retaining structures
experience gained on this contract. The site team made that make up the tunnel complex, and the soil and water
detailed observations of the performance of all 14 conditions in which they were constructed. It sets out the basis
structures during construction, and this paper gives an of the design, and comments on those aspects that affected
overall summary of the wall movements and prop loads work on this contract.
that were measured. The walls were surcharged, and
none of the props was preloaded. Despite this, the Detailed monitoring measurements were made during the
maximum movements were within those estimated construction of all 14 structures (Holmes et al., 2005). The
from published correlations. Wall movements were time paper summarises the wall movements and prop loads that
dependent, and occurred at rates of up to 0.2 mm/day. were measured, and compares these with the design
Prop loads were generally about 40% of the values calculations. Wall movements continued after excavation, and
obtained from moderately conservative calculations. the paper gives the rates of movement that were measured.
Reducing the prop stiffness assumed in calculations
improves agreement, and measurements are reported Prop load measurements are reported that identify the
that provide a basis for closer appraisal of this aspect in important effect of prop stiffness. They show that the prop
future designs. stiffness values that are often assumed in design are unrealistic.
Alternatives are suggested that would improve the accuracy of
NOTATION design calculations.
c9 effective cohesion
E Young’s modulus of concrete
E9h horizontal drained Young’s modulus of soil.
I second moment of area 2. ASHFORD TUNNELS
K0 coefficient of earth pressure at rest CTRL passes through Ashford, Kent in two cut-and-cover
s9 mean principal stress tunnels that, together with approaches and linking earth-
t9 shear stress retaining structures, amount to 1.8 km of contiguous bored
ö9 effective angle of friction piled retaining structures. The tunnels were constructed by
Skanska Construction (UK) in partnership with engineers and
1. INTRODUCTION project managers Rail Link Engineering (RLE) on behalf of
Geotechnical engineers have long recognised the importance of Union Railways.
field observations to provide insight into behaviour
mechanisms, and as a guide to the selection of design Figure 1 shows the common features of the embedded retaining
parameters (Burland et al., 1979). The cut-and-cover tunnels at walls at Ashford. The tunnel site sloped down from north to
Ashford were carefully monitored during construction, south, so that preliminary excavation to the level of the tunnel
providing the opportunity to compare the design of embedded roof generated greater surcharges on the north walls than on
retaining walls with their performance. the south. The walls were formed with contiguous bored piles
of between 900 mm and 1350 mm in diameter (Roscoe et al.,
The tunnels were designed during the period 1998–2002. 2002). Cementation Foundations Skanska bored the piles dry,
Various different standards were current (BSI, 1994; Highways but added bentonite slurry to stabilise the bores during cage
Agency, 1994; Padfield and Mair, 1984), and further changes installation and concreting. A deep dewatering system using
were being introduced to bring geotechnical design within the ejector wells was installed to control the water pressures in the
Eurocode framework (BSI, 1997). A consistent approach was underlying Weald Clay.
Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine 279
The tunnel complex was divided for design purposes into 14 3.3. Design parameters
different structures. The designs generally adopted moderately conservative peak
values for soil properties (Table 3). The strength parameters for
Figure 2 gives the local names of the structures, the method of the Atherfield and Weald Clays were derived from the triaxial
construction that was used, and the extent of deep dewatering test results plotted in Figures 3 and 4. Angles of friction were
in the Weald Clay, and Table 1 gives the key dimensions of the increased by 28 to reflect the difference between triaxial and
walls. plane-strain stress states.
Box names: Chart Advance Chart Road Maidstone Greensands Way Gasworks Lane Cattlemarket Beaver
Road to Maidstone Railway to Beaver Road
Railway Road
Prop’d
Chainage
88 ⫹ 700 88 ⫹ 900 89 ⫹ 100 89 ⫹ 300 89 ⫹ 500 89 ⫹ 700 89 ⫹ 900 90 ⫹ 100
Scale
0 100 200 m
280 Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine
Beaver
Road
7(0)
0(0)
3.8
1.7
0
0
4. GROUNDWATER
Groundwater level prior to construction was between 1 m and
Retained cut
2 m below ground surface. Water levels in the Hythe Beds
Cattlemarket Lane to
5(0)
0(0)
2.3
Beaver Road
0
reduced to base of capping beam level by site drainage.
1.5(0)
12
4
6
wells installed by WJ Groundwater Limited. The extent of deep
dewatering is indicated in Figure 2. Permanent gravity wells
were provided to relieve pore pressures beneath the base slabs
Gasworks
4.5(0)
once the ejector system was removed, and the design water
Lane
10.5
0(0)
5.8
9.5
13
7.7
(8)
10
13
1
13
2
3.2
(8)
10
13
3
0
Greensands
5. CONSTRUCTION METHODS
Way
(8)
10
existing road and rail links that could not be severed and were
10.2
5(7)
4B
12
0
10(7)*
10.2
6(7)
2.1
11
0
14–16
5.9(6)
1(0)
13
was cast on falsework after the base slab was complete. The
0
2
Maidstone
Railway
retained cuts at the east end of the complex (Figure 2) used the
same temporary support method.
12–14
5.9(6)
1(0)
13
0
Railway
2.2(0)
13.1
1.4
1.7
14
3.8(0)
7.6
0
2(4)
2(4)
2.5
8
Figure 5.
Retained height, H: m
south: m
total: m
Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine 281
Soil Plasticity index: % ö9: degrees c9: kPa K0 E9h , initial (gradient): MPa
Table 3. Soil properties: moderately conservative values used for serviceability limit state design
150 moment connection was provided between the roof and walls
of the Greensands Way structure.
100 Most probable
(c⬘ ⫽ 10 kPa, φ⬘ ⫽ 24°)
Moderately conservative Temporary tubular steel props were purpose-made for the
50 (c⬘ ⫽ 0, φ⬘ ⫽ 24°)
contract, and Figure 8 shows the prop and waling system. Each
0 prop could carry a constant-temperature load of 5400 kN, with
0 50 100 150 200 250 300 350 400 450
Mean principal stress, s⬘: kPa a further 1895 kN due to temperature.
Figure 3. Triaxial test results: Atherfield Clay The system allowed for multiple use of the props, and both
length and spacing could be adjusted. Props in the four-track
structures (Figure 7) were assembled from two of the props
from the two-track structures (Figure 6) and a shorter make-up
250 piece. Prop spacing could be increased from 4.5 m to 6 m by
extending the walers during installation.
200
Shear stress, t ⬘: kPa
Concrete-filled bags packed the walers from the wall piles, and
150 in situ concrete thrust blocks were cast between props and
walers. The props were not preloaded. Hydraulic jacks were
Moderately conservative
100 (c⬘ ⫽ 0, φ⬘ ⫽ 26°) inserted during prop removal so that the thrust blocks could be
lifted or broken out and the prop loads reduced incrementally.
50 The system proved robust in practice, and rapid to install. In
normal production a gang of four could prop 18 m of tunnel in
0 three shifts.
0 100 200 300 400 500
Mean principal stress, s⬘: kPa
At Beaver Road retained cut the calculated prop loads exceeded
Figure 4. Triaxial test results: Weald Clay the capacity of proprietary propping systems (Loveridge, 2001).
Lower loads were estimated using the distributed prop load
(DPL) method (Twine and Roscoe, 1999), in conjunction with
load measurements from adjacent sections of the work
shows excavation approaching formation level in the (Loveridge, 2001). This justified the use of hired props and
Maidstone Railway box. Ventilation, lighting, cranage saved over £150 000.
restrictions and limited access all add to the cost of top-
down construction. Temporary props were installed at about 400 locations but
none was damaged or displaced by the ongoing construction
6. PROPPING activities. With appropriate site control and up-to-date safe
The structures with permanent concrete props are shown in working practices it is not considered necessary to treat the
Figure 2. Table 4 gives prop dimensions and spacing. The loss of a prop as one of the design cases.
282 Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine
Retained
height
Temporary Temporary assumed
berm berm
Plunge
columns
Figure 6. Bottom-up construction: Greensands Way Figure 7. Top-down construction: Maidstone Railway box
7. DESIGN
The designs followed the procedure developed by RLE for use Structure Prop size: Prop spacing:
throughout the CTRL project. BS 8002 (BSI, 1994) was adopted m m
as the design standard for cantilever and singly propped
embedded walls. Many of the structures supported road and Chart Road to Maidstone Railway 13 2.3 6
rail loads, and the procedure included additional steps to show Greensand Way propped cut 13 1 6
Gasworks Lane 13 1 4.5
compliance with BD 42/94 (Highways Agency, 1994). The Cattlemarket to Beaver Road 13 1 4.5
Ashford walls fall outside the scope of these documents, and
further reference was made to Eurocode 7 (BSI, 1997) and to Table 4. Permanent prop dimensions
other current developments. This approach has since been
Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine 283
Twin
610 ⫻ 305 ⫻ 238 UB
Waler
Waling
Detail ‘A’ beam Concrete
bags
750 ⫻ 950 in situ
concrete thrust block
Twin
254 ⫻ 254 ⫻ 167 UC
Extension waler 35 mm thick
end plate
Jacking points
for prop removal
0 5m
Scale
The SLS analyses used the design values of soil properties Stage of analysis Bottom-up Top-down
(Table 3) without modification, and the structural capacity of construction construction
the piles was determined by calculating crack widths.
During construction 0.7EI 1.0EI
ULS was checked in two ways. Long term 0.5EI 0.5EI
(a) The bending moments and shear forces from the SLS Table 5. Derivation of wall stiffness
analysis were multiplied by 1.35.
284 Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine
(a) SLS: 15 cases Wall movements for the top-down and permanent propped
(b) ULS determined from an SLS analysis: two cases structures are generally close to the 0.15%H line, with an upper
(c) ULS analysis: three cases bound at about 0.2%H. Results from the structures that
(d ) ULS and SLS equally critical: one case. included an initial cantilever stage are sensibly bounded by the
0.4%H line. The Greensands Way structures were constructed
Analysing each of the walls for both limit states increases using the observational method to utilise more of the walls’
design costs, but this distribution shows that the most critical capacity, and movements were proportionately greater than
case could not be identified in advance. elsewhere. Larger movements were recorded at inclinometer CR
IC1 in the Chart Road box and at the utilities bridge. These are
Once the governing limit state was clearly established, explained by a long delay in construction, and inundation
subsequent amendments to the construction sequence were during regional flooding.
justified for that limit state only.
The Ashford walls support significant surcharges (Figure 2),
7.5.2. Wall stiffness. The uncracked wall stiffness was used to and at Gasworks Lane they retain up to 5.8 m of overburden
analyse the top-down structures, but when excavation was rather than stiff clay. None of the props was preloaded during
complete, flexural cracks were seen at about the level of the installation. Despite these factors, the wall movements lie
maximum bending moment. Clearly, the vertical load on the within the expected range.
walls had not been sufficient to prevent cracking. Using the
cracked modulus in the analyses would have reduced the Figure 12 summarises the ratios between measured and
calculated bending moments and increased the calculated wall calculated movements. Measured movements were about 40%
movements. The temporary props could have been set at lower of those calculated using a moderately conservative approach,
levels to accelerate excavation and reduce costs. but for finite-element analyses were grouped about the
expected value of 70%. The ratios for ‘most probable’ analyses
7.5.3. Relaxation. Frew can model the relaxation of concrete, cover a wide range, but only rarely approached 100%. Many of
taking account of the reduction in bending moment that results these analyses were critical only for the initial cantilever stages
from the lower long-term stiffness (Gaba et al., 2003), but of the excavation.
initial analyses for the Gasworks Lane propped cut did not use
this facility, overestimating long-term moments by about 30% 8.2. Time-dependent movement
of their ‘relaxed’ values. This led to an unnecessarily slow and It has long been recognised that ground movements around
cautious approach during the first excavations. Subsequent excavations in stiff clay have a significant time-dependent
designs took benefit from relaxation to reduce calculated long- component (Burland et al., 1979). This aspect is particularly
term moments. important where an observational method is being used and
controls are set on the rates of movement (Holmes et al., 2005),
Concrete starts to relax from the time that load is applied. This and at Ashford results from the first structures to be
effect is allowed when considering long-term cases, but may constructed were used to optimise the subsequent excavations.
also influence the temporary stages. Design of the Maidstone
Railway box allowed 25% relaxation in the later stages of Wall profiles at different stages of construction (Figures 9 and
construction following prop removal. 10) show that movement continues during periods when there
Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine 285
Temporary prop
Temporary
5 prop 5
Base slab
Formation
10 Key dates
10
Depth: m
Base slab
Formation 15 Dig below prop 30 May 2000
15
Dig to formation 5 June 2000
Key dates Cast slab 21 Toe of wall
June 2000
Depth: m
Excavate to formation
15 May 2001
25
Toe of wall
25
Cast base slab 28 July 2001
Figure 9. Wall movement profiles: top-down construction, Figure 14 summarises all the movements that were recorded
Maidstone Railway box between the end of excavation and base slab completion. Rates
of movement were generally less than 0.2 mm per day, and
Chart Road
Advance
60 Chart Road/Maidstone Railway
Maidstone Railway
Greensand Way – propped
Maximum movement at dig to formation stage: mm
50 Greensand Way – 1 to 4A
bridge
4
Gasworks Lane
0·
(2 props) (2 props)
20
Utilities
bridge
10
5 mm
0
0 5 10 15 20
Retained height, H: m
286 Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine
Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine 287
50 days
20
0·1 mm/day
10 0·2 mm/day
0
04-01-01 23-02-01 14-04-01 03-06-01 23-07-01 11-09-01 31-10-01
Date
25
Chart Road
Advance
20 Chart Road/Maidstone Railway
Maidstone Railway
Greensand Way – propped
Time dependent movement: mm
Greensand Way – 4B
15
Greensand Way – 1 to 4A
y
Gasworks Lane /da
mm
Cattlemarket to Beaver Road 0·2
10
90% consolidation
/day
2½ days 0·1 mm
0
0 10 20 30 40 50 60 70 80
Time: days
Utilities
⫺5 bridge
4000
Cast
Excavate base slab
to formation
3000
Load: kN
2000
1000
288 Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine
0%
10
Greensand Way – 4B
Greensand Way – 1 to 4A
5000 Gasworks Lane
Greensands Way Cattlemarket to Beaver Road
calculations (x) used %
Measured load: kN ‘most probable’ soil 70
properties Calculated using 50%
4000 theoretical stiffness
3000
40%
2000
Utilities
bridge
1000
0
0 1000 2000 3000 4000 5000 6000 7000
Calculated load: kN
Figure 16. Measured and calculated prop loads. Results for 36 sets of props (117 props in total)
Prop location Prop type Measured load: kN Calculated load: kN Measured load/
calculated load: %
installation. Despite this, the overall wall movements fall Wall movements continued after excavation at rates of up to
within the conventional relationships used to estimate wall 0.2 mm/day. This is not explained by consolidation of the stiff
movements. clay soils. Further research is needed into the effects of soil
creep and concrete relaxation.
Measured movements at the end of excavation were only about
40% of the values calculated on a moderately conservative The rates of movement measured on the first structures to be
basis. This ratio increased to around 70% where finite-element constructed were used to optimise the construction sequence
analyses were carried out. for the excavations that followed.
Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine 289
290 Geotechnical Engineering 163 Issue GE5 Design and performance of retaining walls Roscoe • Twine