Investigation of Collapse of Slurry Trench in
Underground Metro Construction
Makarand G. Khare Principal Engineer, AECOM India Pvt Ltd, Chennai, India
M Ramanathan Associate Director, AECOM Asia Co Ltd, Chennai, India
Introduction
The diaphragm wall installation is a first step in construction of a typical cut-and-cover metro station.
The diaphragm walls for underground metro station are required to be constructed with a minimum
intrusion beyond the station footprint because of the limited land availability in congested urban
landscape. The slurry trench technique is the most widely used construction method for diaphragm
walls. In this technique, a trench is excavated and simultaneously filled with bentonite/polymer slurry.
A carefully excavated slurry trench results in little or no soil disturbance (Parkinson and Gilbert, 1991).
However there are reports of ground subsidence or even severe cave-in of slurry filled trench
excavations (Hajnal et al. 1984 and Filz et al. 1997). The trenching process disturbs the in-situ state
of equilibrium in soil mass resulting in inward movement of the trench face. The slurry filled trench is
subjected to lateral earth pressure which includes the effects of surcharge load (due to construction
and/or vehicular traffic and adjacent buildings) and groundwater pressure. The excessive lateral
pressures can lead to collapse of the trench faces. The instability of slurry trench can also threaten
the safety of nearby structures. Therefore, the stability of slurry trenches during construction of
diaphragm wall is of paramount importance in sites where water table is close to the ground surface,
soil is weak, and structures generating high surcharge loads are in close proximity.
Two cases (referred to as Case 1 and Case 2) of slurry trench collapse during the construction of
underground metro stations are discussed in this paper. A polymer based artificial slurry was used in
Case 1 and bentonite slurry was used in Case 2. The stations referred in Case 1 and Case 2 were
constructed by different Contractors.
In case of a trench supported by bentonite slurry, the slurry permeates into soil fabric and forms an
impermeable thin film (also called as filter cake) along the wall. The impermeable film helps to
transfer the hydrostatic pressure on the trench face and gives the stability to the trench. The shearing
resistance of the slurry saturated zone and the plastering effect of the filter cake formed on the trench
face also contributes towards the trench stability.
The polymer based artificial slurry behaves as a pseudo-plastic fluid and unlike bentonite slurry,
achieves the trench support without forming a filter cake (Kaul, 2010). The polymer based slurries
form a membrane on the trench face which enhances the stabilization by transferring hydrostatic
pressure to the walls of excavation. The high concentration of negative charges on soil particles
helps the binding of the soil with the positive charge present on long carbon chain of polymer fluids.
The bond between soil and polymer gives additional stability to the trench.
Site Geology
The regional geology is characterised by an easterly sloping coastal plain. The eastern and southern
parts of the area have shallow bedrock, while the central and northern parts have zones of deep bed
rock. The greater part of the district is covered by Pleistocene to Holocene alluvium of rivers in the
north and in the south. The thickness of alluvium varies from a few meters in southern parts to more
than 50m in central and northern parts. The average depth to bedrock in the district varies from 20m
to 25m. The alluvium consists of silty sand, sandy clay, silt and occasional gravels and is underlain by
crystalline rocks. The natural ground level is about +6.0 m MSL (mean sea level). The ground water
table is shallow and rises close to ground level during rains.
Figure 1 shows diaphragm wall panel layout, and locations of boreholes, collapsed panel, ground
settlement markers and piezometers. The station box consists of two end shafts (used for launching
and retrieval of tunnel boring machines) which are about 3m wider than the middle portion of the
station. Table 1 shows the geological profiles for Case 1 and Case 2. The geological profiles are
deduced from nearby borehole and the panel borelog. Figure 2 shows the distribution of standard
penetration test blow count (SPT ‘N60’) and the percentage of fines (i.e. silt and clay) content obtained
from the sieve analysis of disturbed samples collected from nearby boreholes.
(a)Case 1
(b)Case 2
BH- borehole, GSM- ground settlement marker
Figure 1 Panel layout, borehole locations, piezometer and ground settlement marker locations
Table 1. Geological Profile
Case No. Depth below ground level (m) Description of Stratum
0 to 2.5 Made ground, pavement
2.5 to 8.5 Silty Sand
8.5 to 10 Silty Sandy Clay
Case 1 10 to 22 Silty Sand
22 to 24 Silty Sandy Clay
24 to 27 Completely Weathered Rock
> 27 Fresh Rock
0 to 4 Low to high plastic Clay
Case 2 4 to 17 Silty Sand
17 to 28 Low plastic Clay
SPT 'N60'
Fines (%)
0 10 20 30 40 50 60 0 10 20 30 40 50 60 70 80 90
0 0
Case 1
5 5
Depth below ground (m)
Depth below ground (m)
Case 2
10 10 Case 1
Case 2
15 15
20
20
25
25
a) SPT ‘N60’ with depth b) Fines (silt and clay) % with depth
Figure 2 Engineering properties of soil
Case 1
The collapse of a 5m long and 1m thick diaphragm wall panel with toe depth of 25.2 m below ground
level is described here. Figure 1 (a) shows diaphragm wall panel layout, and locations of boreholes,
collapsed panels, ground settlement markers and piezometers. The trench excavation was carried
out using hydraulic grab. The panel was being excavated with two 2.5 m long cuts. The first cut was
excavated to the design depth of 25.2m. In the second cut, the weathered rock was encountered at a
depth of 24.8m below ground. The weathered rock could not be excavated with grab and therefore a
sharp edged drop weight was used to chisel the rock. The chiselled rock was then removed by the
grab. The completion of panel was delayed because the rock chiselling was hampered during heavy
rains.
This station was being constructed on a busy road. Therefore, the diaphragm wall reinforcement
cage fabrication was carried out in an open barricaded area on the road. Due to heavy rains the
welding of reinforcement cage was deferred. A public holiday further delayed the completion of the
trench. On the sixth day (after the start of trench excavation) when the work resumed after holiday,
the trench was found to have collapsed at a depth of about 7 m below ground. The contractor carried
out KODEN test up to depth of 6.3m in the first cut and 9.5m in the second cut. The results of
KODEN test carried out on the second day of excavation and that after the collapse (on the sixth day)
are shown in Figure 3.
The KODEN test is based on the principle of ultrasonic wave travel from sensor to the wall of trench
and reflection of the wave from trench wall back to the sensor through bentonite/polymer slurry media.
The time taken by the wave to travel in the slurry and the wave velocity is then interpreted to assess
the geometric profile of the trench. The vertical line (A) shown in Figure 2 represents the vertical
descent path of the probe. The trench wall profile is shown by the inner borders of dark strips (line B).
Figure 3 shows the collapse at about 1m below ground and extending to a depth about 9.5m. The
collapse occurred on the excavation face (EF) of the diaphragm wall (i.e. the inner face of
underground station).
B A B
* #
1.0 0.5 EF 0 SF 0.5 1.0
0 0 0 B * A # B
1.0 0.5 EF 0 SF 0.5 1.0
5 0
Collapsed
10 Trench
5
15
20 *Excavation Face
#Soil Face
DEPTH 9.5 m
DEPTH 23.8 m Hm m
a) Before collapse Hm m
b) After collapse
Figure 3 KODEN test results – Case 1
The grain size analysis of samples obtained from the nearest borehole show loose silty sand with SPT
‘N60’ generally less than 10 and with fines content less than 10% (as shown in Figure 2) upto a depth
of about 10m below ground level. The low fines content could have possibly reduced the membrane
like effect of slurry which controls the fluid loss and helps in building significant positive hydrostatic
pressure from slurry to the excavated face of the trench. The heavy rains caused the rise in water
table close to a depth of 2.5m below ground level resulting in reduction of net positive slurry head and
the factor of safety for upper 10m of loose sandy soil. The polymer manufacturer recommended
3
slurry unit weight of 10 to 10.4 kN/m . The polymer slurry viscosity was maintained at 54 seconds
3
and the slurry unit weight was maintained at about 10 kN/m at the time of collapse, i.e. at the lower
end of the recommended range. The factor of safety for the upper 10m thick loose sand layer at the
time of collapse is estimated using the approach suggested by Schneebeli (1964). Schneebeli (1964)
investigated the effect of arching on short panels on the basis of the theory of lateral earth pressure
on silos (Caquot and Kerisel, 1956). In this method, the factor of safety is obtained by comparing the
lateral earth pressure (with allowance for arching effect) diagram with the net fluid pressure (slurry
pressure minus groundwater pressure) diaphragm. In case of cohesionless soils (c’=0, ’ soils) the
factor of safety at the water table level and at the base of the trench is obtained using following
expressions:
At water table level, factor of safety (F) = --------------------------------------(1)
At base of trench, factor of safety (F) = --------------------------------------(2)
Where,
n = z/L
f = unit weight of slurry
w = unit weight of water
Hf = depth of slurry
Hw = depth of water
z = depth of soil layer analysed
L = length of trench
d = dry unit weight of soil above water table at a depth z
sub = submerged unit weight of soil below water table at a depth z
’
= effective friction angle of soil
’ ’
N = flow factor = (1+sin ) / (1-sin )
The slurry properties and shear strength of soil used for the analysis are given in Table 2. The Figure
4 shows the factor of safety calculated at various depths using equations 1 and 2. The depth of water
table is considered at 2.5m below ground level as recorded by the nearby piezometer. The slurry unit
3
weight at the time of collapse is taken as 10.0 kN/m . It is evident that the factor of safety was less
than 1.0 when water table was just 2.5m below ground. The Schneebeli (1964) approach accounts
for the arching effect. In case 1, the trench was open for 6 days, therefore the beneficial effect of
arching could have reduced with time. The creep effect could have reduced the shear strength of
upper loose sand than that assumed in the back analysis. Therefore, the actual factor of safety at the
time of collapse could have been less than those shown in Figure 4.
Table 2. Parameters used for back analysis
Property Case 1 Case 2
3
Bulk weight of soil (kN/m ) 17 18
3
Unit weight of slurry (kN/m ) 10, 10.2 10.4, 10.9
Effective friction angle of soil ’ (degrees) 30 32
The contractor implemented following measures to increase the factor of safety:
3 3
a. The unit weight of slurry was increased from 10 kN/m to 10.2 kN/m to increase the slurry
pressure.
b. The slurry head was maintained at about 3m above water table. For this purpose a 0.5m high
bund made up of sand filled bags was created around the trench as shown in Figure 5. The
bund helped to keep a higher positive slurry head when water table was close to ground level.
c. The reinforcement cage was made ready before the start of trenching to reduce the cycle time
of diaphragm wall panel construction.
The factor of safety is again estimated to reflect the corrective measures. The factor of safety
3
estimated using higher slurry unit weight (of 10.2 kN/m ) and that considering a higher slurry head as
well as unit weight are shown in Figure 4. From Figure 4, it is observed that the increase in the slurry
unit weight alone resulted in a marginal increase (of less than 8%) in the factor of safety. The
increase in the slurry head combined with increased slurry unit weight resulted in about 25% increase
in the factor of safety. In this case, the increase in slurry head played major role in increasing the
factor of safety of trench. It is evident that these simple and inexpensive measures resulted in
substantial increase in the stability of the top 10m of loose sandy soil. A large number diaphragm wall
panels were completed successfully by implementing above measures.
Figure 1 show a settlement marker (GSM 01) installed on the pavement and very close to the
collapsed trench. The settlement marker did not show any movement as a result of trench collapse.
A membrane like effect provided by the upper crust of pavement appears to have spanned over the
collapsed soil. The relaying of road over the older paved surface (during maintenance by the
municipal corporation) may have resulted in strong yet flexible upper crust with a limited capacity to
span over the cavity.
1.4
Af ter increase in slurry unit weight and head
1.35
Af ter increase in slurry unit weight
1.3
At the time of collapse
1.25
Factor of Saf ety
1.2
1.15
1.1
1.05
1
0.95
0.9
0.85
2 3 4 5 6 7 8 9 10
Depth of soil layer below ground (m)
Figure 4 Estimated factor of safety – Case 1
A bund of sand filled bags
Figure 5 Bund to increase the positive slurry head
Case2
The collapse of 4.5m long and 1m thick diaphragm wall panel with toe depth of 22.6m below ground
level is described in this section. The locations of collapsed panel, boreholes and piezometer are
shown in Figure 1(b). A week long heavy rainfall resulting in the rise of water table was the primary
reason for the collapse of trench in Case 2. The rise in water table reduced the factor of safety in silty
sand observed from a depth of 5 m to 15m. From Figure 2, it is observed that the fines content in this
stratum is about 10% and a typical SPT ‘N60’ value is about 15. The records of ground water levels
during the trenching and immediately after the collapse are not available. The available groundwater
level record shown in Figure 6 indicates that at the time of trench collapse, the groundwater was likely
to have been at a depth of about 2m below ground level. The geological profile given in Table 1 and
the soil parameters given in Table 2 are used to investigate the effect of rise in water table on the
factor of safety. The factor of safety is calculated using Schneebeli’s method described in the
preceding section. The factor of safety for the soil at a depth of 10m below ground (which
corresponds to the middle of upper sandy stratum) is analysed and the results are shown in Figure 7.
Groundwater Monitoring Days
0 5 10 15 20 25 30
0
Depth of ground water below ground
0.5
Trench collapse
1
1.5
(m)
2.5
3.5
Figure 6 Ground water monitoring data – Case 2
3
1.6 slurry unit weight = 10.4 kN/m
1.5 X slurry unit weight = 10.9 kN/m
3
1.4
Factor of Saf ety
1.3
1.2
1.1
1
0.9
0.8
0.7
0.6
0 0.5 1 1.5 2 2.5 3
Depth of water below ground level (m)
Figure 7 Effect of rise in ground water on Factor of Safety – Case 2
From Figure 7, it is evident that when the water table rose from 3m to 2m below ground level the
factor of safety dropped to less than 1.0. The maximum water table recorded was just 1.5 m below
3
ground. At this site, the contractor decided to increase the slurry unit weight from about 10.4 kN/m to
3 3
about 10.9 kN/m (which corresponds to the higher end of the range of 10.4 to 11.0 kN/m
3
recommended by Contractor’s method statement). The increased slurry unit weight (of 10.9 kN/m )
could counter the high hydrostatic pressure even when the water table rose to about 1.5m below
ground level as is evident from Figure 7. From Figure 7, it is observed that the increase in slurry unit
weight improved the factor of safety by about 31% and 14% when water table level was at 1m and 3m
respectively. The deeper water table results in increase in the lateral earth pressure and reduces the
beneficial effect of increase in slurry unit weight on the factor of safety.
Increasing the unit weight of slurry beyond a certain value could hamper the displacement of slurry by
concrete. The bentonite concentration is usually kept in the range of 4-8 percent (which represents
3
the slurry density in the range of about 10.2 to 10.4 kN/m ) so that the desired rate of flow of concrete
can be maintained without compromising on the formation of filter cake on the excavated face. The
rising concrete during tremie concreting may not be able to dislodge the filter cake from the trench
face (if the slurry weight is high) and reduce the skin friction. Therefore it is prudent to adopt a
combination of measures which may include increasing the unit weight of slurry and increasing the
slurry head in order to improve the factor of safety without compromising the performance of
diaphragm wall.
Conclusions
The safety of diaphragm wall trench excavation is of prime importance in underground metro projects.
The stability of slurry trench can be compromised due to unfavourable geological profile or rise in
water table. Two cases are described where the collapse of a slurry trench was observed. The factor
of safety for the collapsed trench was back-analysed to investigate the cause of failure. The primary
factors responsible for the instability of slurry trench were the rise in water table which lead to
reduction in positive slurry head and prolonged trenching operations due to poor construction
planning. The risk of trench collapse appears to be independent of the type of slurry (i.e. bentonite or
polymer). The corrective actions were aimed to increase the margin of slurry pressure over the earth
pressure acting on the trench wall. A bund using sand filled bags was created up to 0.5m above
guide wall top so that slurry head can be maintained at least 2.5m above water table. The
construction activities were planned to minimise the overall time cycle for the diaphragm wall panel
construction. A combination of measures such as increasing the slurry unit weight and slurry head
were used to improve the factor of safety by about 25%. The measures to prevent the collapse of
slurry trench require very little resources and cost in contrast to the consequences in the event of
collapse.
References
rd
Caquot, A. and Kerisel, J. (1956). Traite de Mecanique des Sols, 3 Edition, Gauthier-Villars, Paris.
Filz, G. M., Boyer, R. D., and Davidson, R. R. (1997). ‘‘Bentonite-water slurry rheology and cut-off wall
trench stability.’’ Proc., In Situ Remediation of the Geoenvironment, GSP 71, ASCE, Reston, Va.,
139–153.
Hajnal, I., Marton, J., and Regele, Z. (1984). Construction of diaphragm walls. John Wiley & Sons,
Inc., New York.
Kaul Krishan (2010). Cut-and-Cover Metro Structures Geo-Structural design: An Integrated
Approach, Spon Press.
Parkinson, J.J., and Gibert, C.M. (1991). Design and construction of slurry walls. In Proceedings of
the American Society for Testing and Materials International Symposium, Atlantic City, N.J., pp. 27–
28.
Schneebeli, G. (1964). ‘Le Stabiliti des tranchees profondes forees en presence de boue’, Houille
Blanche, 19 (7), 815-820.
