Numerical Methods in Geotechnical Engineering – Schweiger (ed.

)
© 2006 Taylor & Francis Group, London, ISBN 0-415-40822-9

Numerical investigation of the failure of a shotcrete lining

R. Pöttler
ILF Consulting Engineers, Rum/Innsbruck, Austria

ABSTRACT: In the course of a 7-km-long railway tunnel, which is situated mainly in weathered chalk marl,
covered by Tertiary over a length of about 500 m, the tunnel traverses a transition zone of chalk marl, weathered
chalk marl and Tertiary. In a specific area with a length of 100 m, the deformation during tunnel drive did not cease
and exceeded the predicted values considerably. About one year after the installation of the primary lining, cracks
occurred in the shotcrete lining of the invert. Many different geotechnical models have been discussed to explain
this special ground behaviour and the failure. Numerical calculations based on extensive geotechnical investiga-
tions were performed, aiming at detecting the cause of this failure. It was only when the real ground behaviour and
reasons of failure had been understood that the secondary lining could be designed in an optimised way.

1 PROJECT shotcrete lining of the heading was situated on rock

mass. The FE model used for calculation is shown in
The tunnel is constructed according to NATM princi- Figure 1. The material law of Tertiary and of the rock
ples. Excavation of heading is followed by excava- mass is Drucker Prager. Input parameters are listed in
tion of bench and invert. Between shotcrete lining and Table 1. Differentiation in the chalk marl (TER1 to
cast-in place inner lining a waterproofing membrane is
installed. A drainage system will prevent any water
pressure from acting on the lining. The inner lining has
to take over the loads resulting from the traffic in the
tunnel, the loads of the electrical–mechanical installa-
tions of tunnel and of the deterioration of shotcrete, as
well as the loads of the primary shotcrete lining and
of the anchors which will corrode with time.
In the area discussed, the tunnel is situated in the
transition zone of Tertiary (TER) and weathered rock
mass (RM) of chalk marl. The interface was predicted
to be monotonous, both in longitudinal and transverse
direction. Numerical analyses for verifying the stabil-
ity of the tunnel were done prior to the construction
based on the geotechnical report. The geotechnical
parameters were verified and improved by back
analyses. The main input parameters, for the finite
element calculations, modulus of elasticity E, and the
undrained parameters cohesion cu and angle of fric-
tion wu could be determined quite well. Supervision
of the tunnel drive was done by comparing and inter-
preting the geotechnical measurements on site with
the calculated displacements.

2 EXCAVATION OF HEADING, BENCH

AND INVERT

In the area discussed the overburden is about 25 m

thick. According to the prognosis, the foundation of the Figure 1. Numerical model for FE calculation.

313
Table 1. Geotechnical parameters for numerical analyses
prior to excavation.

E Load E Unload cu wu g
[MN/m2] [MN/m2] [MN/m2] [°] [MN/m3]

TER1 6 18 0.010 25 0.020

TER2 16 48 0.015 25 0.021
TER3 80 240 0.015 25 0.021
RM 100 200 0.100 30 0.022

Figure 3. Additional support measures (elephant footing,

vertical anchors).

Figure 2. Observation of displacements.

TER3) takes into account that the weathering of rock

mass nearer to the surface was expected to be higher. Figure 4. Result of c/w reduction.
The thickness of the shotcrete lining was 280 mm.
In addition to the lattice girders, anchors with a length
of 6 m spaced every 2 m ⫻ 2 m were installed. For the
excavation of the heading the following ranges of dis-
placements were calculated (Figure 2):
– Roof settlement (RS) 45 mm–50 mm
– Horizontal convergency (HC) 10 mm
– Settlement of footing (SF) 25 mm–30 mm
As there was no prognosis on long-term behaviour of
the ground it was expected that displacements due to Figure 5. Calculated and measured roof settlement (RS).
tunnel drive would cease 1–2 diameters behind the
face. The normal force in the shotcrete lining was
2.1 MN/m, resulting in stresses of 3.5 MN/m2 in the
footing of the lining. Although these stresses are well
below the sustainable stresses of the rock mass, addi-
tional numerical analyses were performed to investi- is well over the required factor of safety according to
gate the influence of a enlargement of the shotcrete the relevant standard. The difference of the partial
lining at the footing (“elephant footings” – Figure 3) factor of safety in the shotcrete lining FNM (Pöttler
up to 800 mm to reduce the stresses in the footing of 1992) with and without enlargement of the lining was
the heading. This investigation was done by using the less than 5% (Figure 5), FNM being the factor of sus-
c/w reduction method. As can be seen from Figure 4 tainable normal force and normal force. The sustain-
there is only very little influence on the displacements able normal force is calculated taking account of the
of the lining. The x-axis shows the reduction factor bending moment and the shotcrete strength divided
(h), the y-axis shows the roof settlements. Figure 4 by the factor of safety according to the relevant stan-
also shows that the factor of safety h, which can be dards (DIN 1045). Thus in case FNM is higher than
calculated 1.00, there is a safety margin.

314
 at that distance (Figure 6). As the geotechnical param-
eters of the weathered rock mass have been assumed
to be almost the same as for the Tertiary, from a geot-
echnical point of view the ground behaviour could not
be explained. Additional support measures had to be
installed based on engineering judgement to prevent a
collapse of the tunnel. In a 1st step these consisted of
enlarging the shotcrete lining at the footing, in a 2nd
step of installing vertical anchors at the basement and in
a 3rd step a temporary invert of the heading, 5 m wide
and at a distance of 15 m. The 1st and 2nd step did not
have any influence on the deformation behaviour of
the heading. Thus the numerical investigations done at
the beginning, showing no major influence of these
support measures, have been proved. Deformations only
ceased after installing, in the relevant section, a tem-
porary invert arch in a 3rd step. Excavation of bench
Figure 6. Measured displacements. and invert had to be done close to the excavation of
the heading. Also, in the invert a shotcrete lining with
a thickness of 250 mm was installed.
Table 2. Geotechnical parameters for adopted numerical
analyses.

E Load E Unload cu wu g 3 STOP OF CONSTRUCTION WORK –

[MN/m2] [MN/m2] [MN/m2] [°] [MN/m3] ADDITIONAL GROUND INVESTIGATIONS

TER1 6 18 0.010 25 0.020 After the shotcrete invert had been placed, the work
TER2 16 48 0.015 25 0.021 on site was stopped for about 1.5 years. During this
TER3 80 240 0.015 25 0.021 time additional ground investigations were per-
RM 50 100 0.030 25 0.022 formed. As the tunnel was situated in karst, geologi-
cal anomalies due to karst phenomena located in the
Enlargement of the shotcrete lining has only minor vicinity of the tunnel had to be detected, which may
influence on the stress-strain behaviour of the tunnel. have an influence on the permanent stability and ser-
The same investigation was done for the placement of viceability of tunnel for a service life of 100 years.
additional vertical anchors at the footing. This addi- These investigation were done by geophysical meas-
tional support measure also had no major influence urements and drilling in an area of about 15 m around
on the overall stability of the heading. Therefore there the tunnel (Pöttler 2004).
was no reason to carry out both support measures Using these investigations, it was possible to gen-
from an engineering point of view. erate a realistic geological 3D model. The 3D model
In the course of the tunnel driving, the measured of the area under discussion is shown in Figure 7. A
displacements became bigger than the calculated dis- major discrepancy between the geological model in
placements (Figure 6). This was mainly because the the prognosis stage and reality is evident. A main dis-
transition zone of Tertiary and rock mass was located crepancy is that the weathered rock mass is not situ-
lower than in the prognosis. Numerical analyses with ated in the area of the heading but under the invert.
reduced values for the geotechnical parameters (Table The surface of the weathered rock mass is quite irreg-
2) resulted in higher allowable displacements. No dif- ular both in longitudinal and transversal direction.
ferentiation between RM1 and RM2 was done in the
numerical model.
For the excavation of the heading the following 4 UNPREDICTED EVENT
ranges of displacements were calculated:
During this period of no construction, regular inspec-
– Roof settlement (RS) 75 mm–85 mm
tions of the shotcrete lining were undertaken by the
– Horizontal convergence (HC) ⫺10 mm–60 mm
site management. One year after the construction
(- .. divergence)
works had been stopped, cracks appeared in the invert
– Settlement of footing (SF) 65 mm–80 mm
lining and additional settlements occurred (Figure 8).
It was assumed that the displacements would cease The settlements could only be stopped when the tunnel
1–2 diameters behind the face. But the results of the was partly refilled with material and wooden columns
measurements showed no abatement of displacements placed to support the shotcrete lining of the heading.

315
Figure 7. Geological situation in reality.

Due to the large settlements of up to 150 mm, the tun- geotechnical parameters as listed in Table 2 have been
nel lining had to be reconstructed to provide enough used for the numerical analyses. Taking account of the
space for installing the inner lining. Additional excavation sequence, the numerical analyses yielded
drilling had been performed to get even more infor- the same results in terms of displacements which were
mation on the ground in this area. The main questions measured (Table 3), therefore it could be assumed that
to be answered were: the ground behaviour and the excavation process were
modelled realistically.
– What is the reason for the unpredicted restart of
But the numerical analyses showed no failure of the
movements?
shotcrete lining in the invert section. The stresses in the
– What is the realistic load on the inner lining caused
shotcrete lining were considerably lower than the sus-
by shotcrete deterioration?
tainable stresses. The actual location of the geological
Different scenarios to explain the behaviour of the strata differed from the prognosis, but this could not
tunnel have been discussed: be the reason for the unpredicted event.
Scenario 1: Change of location of geological strata: Scenario 2: Influence of rock “peak”: As can be
The tunnel is situated in Tertiary and the weathered seen from Figure 4, the weathered rock mass forms a
rock mass is about 10 m below the invert. The same peak in the vicinity of the area where the crack in the

316
Figure 8. Cracks in the shotcrete lining of the invert.

Figure 9. Numerical model for FE calculations.

Table 3. Comparison of numerical analyses and geotechni-
cal measurements.

RS [mm] HC [mm] SF [mm]

safety. This was in addition proved by geotechnical
Numerical analysis 114 75 100 measurements.
Measurements 125 80 125 Scenario 5: Long-term behaviour of rock mass:
The geological formation of the Tertiary and Chalk
Marl does not show any aspects of creeping or
swelling. The tunnel is situated well above the perma-
shotcrete lining appeared. This geological situation nent groundwater level. No change in groundwater
was also analysed, assuming a great difference in the conditions occurred in the relevant period.
stiffness behaviour between Tertiary and rock mass. The Scenario 6: Pore water pressure: The change of
modulus of elasticity of the rock mass was assumed to stresses and strains due to the excavation process also
be 300 MN/m2. The influence of this rock “peak” on the changes the pore water pressure and thus the effective
stresses in the shotcrete lining was minor. No failure of stresses of the ground. In case of low permeability
shotcrete lining occurs. this process takes a long time. The permeability of the
Scenario 3: Influence of the rock “peak” and addi- Tertiary is high so that this process, even if it had had
tional weathering of ground in the invert. It is assumed occurred, would have taken only a few days. From
that the Tertiary under the shotcrete lining of the invert this point of view, the change of pore water pressure
weathers due to water ingress from inside the tunnel, as the reason for the unpredicted event can be
cohesion decreases to zero and the modulus of elastic- excluded.
ity is considerably lower. This scenario is rather unre- Twelve different possible scenarios with different
alistic as no water was in the tunnel and only a small numerical methods and models were investigated to
wet zone of Tertiary of about 100 mm was detected ascertain the reason for the failure and thus to be able
when the shotcrete lining in the invert was replaced. to answer the two questions above in a satisfactory
The numerical analyses based on this model (Figure manner. Some of these scenarios are more unrealistic
9) showed that a failure of shotcrete lining would than the one described, such as heave of the tunnel
occur only if the weathered zone was very large. due to groundwater and change of temperature. But
Scenario 4: Influence of excavation procedure of the every scenario that was brought up by an involved
heading: The heading was driven without applying an expert was discussed. None of the investigated twelve
enlarged footing of the shotcrete lining. Scenario 4 geotechnical models could answer the question in a
shows that because no enlarged footing was con- satisfactory manner.
structed, the amount of vertical displacement of the In summary, the discussion and investigation
heading was so great that a loosening of the Tertiary at showed that a reason for the unpredicted event could
the crown occurred, resulting in the total load of over- not be identified, not even after a lot of additional
burden resting on the lining. The geotechnical parame- numerical analyses and additional ground investiga-
ters of the Tertiary have been proved and the numerical tions. The dimensioning of the inner lining was there-
analyses based on these parameters made it possible to fore done in a very conservative way. The lining was
model the overall behaviour very realistically. These designed so that it could sustain the entire loads of the
numerical analyses also showed that the influence of a overburden. The thickness of the lining is 550 mm,
enlargement of the footing of the lining has only concrete grade B45 (DIN 1045) and reinforced by
minor influence on the behaviour and on the factor of 23.2 cm2/m on both sides.

317
5 CONCLUSION the process one had to become aware that, for the
present, the limit of understanding and numerical
The prognosis of the geological situation was done modelling has been reached. It becomes evident that
based on the well-known geological mapping and tunnelling in such ground conditions can be done
additional drillings spaced every 500 m over the only by an experienced team. The numerical analyses
course of the tunnel. After problems as described are helpful but not the only tool for success.
appear, “experts” may state that a greater number of There was no complete collapse of the tunnel and
ground investigations have to be carried out before no loss of life. This section of the tunnel was expen-
tunnel driving. It goes without saying that the better sive, but all in all economic with regard to other pos-
the ground investigations are before the construction sible and realistic scenarios. This was thanks to the
starts, the lower the risk of an unpredicted event. The proper work done on site by the client, construction
amount and quality of ground investigation represent- company, site supervision and designer.
ing the state of the art is found in the relevant stan-
dards. But even if the standard is met, like in the
described case study, unpredicted events may occur.
In this case study, the reason for the unpredicted event REFERENCES
could not be defined in a proper way, although a far
greater number of ground investigations were per- Pöttler, R. 1992. Die Standsicherheitsuntersuchung für die
Kaverne der englischen Überleitstelle im Kanaltunnel.
formed than ever done in a tunnelling project and Bautechnik 69: 602–617.
although the geological situation was well known. Pöttler, R. 2004. Die Beherrschung der Karst- und
As numerical analyses were used to a great extent, Erdfallproblematik bei Tunnelbauwerken. Veröffentlichung
and as there was time for a lot of discussion and meet- des Instituts für Geotechnik der Technischen Universität
ings after the completion of the tunnel, at the end of Bergakademie Freiberg. Freiberg.

318