Sie sind auf Seite 1von 11

Fachthemen

Georgios Anagnostou DOI: 10.1002/gete.201200014


Paolo Perazzelli

The stability of a tunnel face with a free span


and a non-uniform support
We investigate the effects of the support pressure distribution and
of the free unsupported span on the stability of the tunnel face by
means of limit equilibrium computations, which are based upon
the so-called method of slices. Design equations and diagrams
are introduced for the estimation of the necessary support pres-
sure taking into account these effects assuming a linear pressure
distribution. Furthermore, the errors introduced by the common
assumption of support pressure uniformity are discussed for the
typical distributions of face support pressure in tunnelling by
slurry shields, EPB shields and compressed air shields. The error
is less than 10 % in the case of closed-shield tunnelling but may
be significant in the case of face reinforcement.
Fig. 1. Failure mechanism and definition of support pressu-
Der Einfluss der Stützdruckverteilung und der freien Spannweite re s(z) and of unsupported span e
auf die Stabilität der Tunnelortsbrust. Im vorliegenden Beitrag Bild 1. Bruchmechanismus und Definition von Stützdruck
wird untersucht, in welchem Maß sich der erforderliche Stütz- s(z) und der nicht gestützten Abschlagslänge e
druck im Vorhandensein einer freien Spannweite zwischen Aus-
bruchsicherung und Ortsbrust erhöht. Ferner wird der Effekt der
Verteilung des Stützdrucks unter der Annahme eines linearen Unstable face conditions necessitate the application
Verlaufs quantifiziert. Das zugrunde liegende Berechnungsmodell of auxiliary measures such as ground improvement,
untersucht das Grenzgleichgewicht eines Bruchmechanismus drainage and face support. The pressure exerted by the
nach der sogenannten Lamellenmethode. Die Berechnungs- support is in general not constant over the face. This is
ergebnisse werden in Form von Bemessungsformeln und Dia- true even in the case of face stabilization by uniformly dis-
grammen dargestellt. Ferner wird der Fehler der gängigen verein- tributed face bolts and also in most cases of closed shield
fachenden Annahme eines gleichmäßig verteilten Stützdrucks für tunnelling (Fig. 2):
typische Problemstellungen diskutiert. Der Fehler beträgt weni- – The simplified assumption of uniform face support pres-
ger als 10 % im Fall von geschlossenen Schildvortrieben (Erd- sure is strictly true only in the theoretical case of com-
druck-, Flüssigkeits- oder Druckluftstützung), kann aber im Fall pressed air application above the water table (Fig. 2a).
einer Ortsbrustbewehrung sehr groß und auf der unsicheren Below the water table (Fig. 2b), a part of the air pressure
Seite sein. just compensates the pore pressure. Only the air pres-
sure in excess of the pore pressure (hereafter referred to
1 Introduction as “excess pressure”) represents a support pressure.
Since the air pressure is constant over the face but the
The present paper deals with the effects of the unsupport- pore pressure increases linearly with depth (hydrostatic
ed span e and of the support pressure distribution s(z) on distribution), the excess pressure is not uniform but de-
the stability of the tunnel face (Fig. 1). An unsupported creases linearly from the crown to the floor of the tun-
span between the temporary lining and the face exists al- nel.
most always in conventional tunnelling during the critical – In slurry-shield tunnelling above the water table (Fig.
period between the completion of the excavation step and 2c), the support pressure increases linearly from the
the support application. In poor quality ground the unsup- crown to the floor according to the unit weight of the
ported span is typically 0.75–1.5 m wide, but can be re- slurry (γs). Below the water table (Fig. 2d), a part of the
duced to zero with pre-support by spiles or forepoling. An slurry pressure is spent to balance the water pressure.
unsupported span may exist also in mechanized tun- The excess pressure increases, therefore, linearly with
nelling, for example in the case of open type gripper TBMs depth according to the difference between the unit
or even of shield tunnelling, if the TBM is operated open- weights of the slurry and of the water (γs – γw). As the
mode and the ground does not close the gap around the slurry density is in general low (1.1–1.2 kg/l, depending
shield. on the bentonite content and on the amount of dis-

40 © 2013 Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · geotechnik 36 (2013), Heft 1
G. Anagnostou/P. Perazzelli · The stability of a tunnel face with a free span and a non-uniform support

(a) (b)

(a) (b) Fig. 3. (a) Face reinforcement by sufficiently long bolts, (b),
support pressure distribution given by the bolts
Bild 3. (a) Ortsbrustverstärkung mit genügend langen An-
kern, (b), Verteilung des durch die Anker ausgeübten Stütz-
drucks

increases linearly from zero (at the tunnel floor) to its


maximum value (at the tunnel crown), the support pres-
sure is also linearly distributed (Fig. 3b). The maximum
support pressure

(c) (d)
s(H) = nH tanω πdτm, (1)

where n is the density of the face reinforcement (bolts per


unit face area); H is the face height; ω is the angle between
sliding plane and face; d is the grouted bolt diameter; and
τm is the bond strength between grouted bolt and sur-
rounding soil [1].
There are many publications dealing with the assess-
ment of tunnel face stability. Readers are referred to the
reviews, e.g., of Idinger et al. [2], Mollon et al. [3] and Per-
azzelli and Anagnostou [4]. The effect of the unsupported
(e) (f)
span was included in previous limit equilibrium models
Fig. 2. Support pressure distributions (grey = excess pressu- [5], [1] and in the computational model of Veermer and
re): compressed air, (a), above or, (b), below water table; Ruse [6], which is based upon numerical stress analysis. In
pressurized slurry, (c), above or, (d), below water table; earth a recent work, Li et al. [7] investigated by means of numer-
pressure balanced shield, (e), above or, (f), below water table ical stress analyses the stability of the face for a particular
Bild 2. Stützdruckverteilung (grau = Drucküberschuss):
case of slurry shield tunnelling with non-uniform support
Luftdruck, (a) über oder (b) unter dem Wasserspiegel; Druck-
schlamm, (c) über oder (d) unter dem Wasserspiegel; erd- pressure distribution.
druckgestützter Schild, (e) über oder (f) unter dem Wasser- The computational model introduced in this paper
spiegel represents a more general version of the method of slices
[8], which approximates the tunnel face by a rectangle (of
height H and width B) and considers a failure mechanism
persed excavated material), the pressure gradient (γs – that consists of a wedge at the face and of the overlying
γw) is rather small and the support pressure distribution prism up to the soil surface. The geometry of the prism
almost uniform. takes into account the possibility of an unsupported span
– Similar remarks apply to the case of EPB shields (Fig. 2e between lining and face (e in Fig. 1). In analogy to silo the-
and 2f). Assuming hydrostatic pressure distribution of ory, the method of slices assumes proportionality between
the excavated material in the working chamber (i.e., ne- the horizontal stress σy and the vertical stress σz:
glecting the shear strength of the muck), the support
pressure gradient can be taken equal to the unit weight σy = λσz (2)
γm of the muck (above the water table) or to the differ-
ence γm – γw (below the water table) and is in general where λ is a constant (coefficient of lateral stress). In or-
higher than the gradient of slurry shields. der to calculate the distribution of the vertical stresses σz
inside the wedge, one considers the equilibrium of an in-
Concerning the support pressure distribution in the case finitesimally thin slice (Fig. 4). The method of slices repre-
of face reinforcement, consider the simplified case of uni- sents an improvement of the model of Anagnostou and
formly distributed bolts. If the bolts are sufficiently long Kovári [9] in that it eliminates the need for an a priori as-
and have a sufficiently high tensile strength (Fig. 3a), the sumption of the distribution of the vertical stress σz in the
anchorage length inside the failure wedge represents the wedge and offers the possibility of analyzing cases with
limiting factor for the bolt forces. As the anchorage length non-uniform face support and heterogeneous ground con-

geotechnik 36 (2013), Heft 1 41


G. Anagnostou/P. Perazzelli · The stability of a tunnel face with a free span and a non-uniform support

Fig. 4. Forces acting upon an infinitesimal slice.


Bild 4. Kräfte, die auf eine infinitesimale Lamelle wirken

sisting of horizontal layers. In the present paper, we main-


tain the assumption of soil homogeneity made in [8], but
consider a linearly distributed support pressure over the
height of the face:

 z 
s = s0 + δs  − 0.5 , (3)
H 

where s0 denotes the support pressure at the tunnel axis


and δs is equal to the difference between the support pres-
sure at the crown and the support pressure at the floor, i.e.
δs = s(H)–s(0).
The computational predictions of the method of Fig. 5. Normalized support pressure s/γ D of a granular
slices agree very well with published results of small-scale material (c = 0) and of a cohesive frictional material as a
centrifuge- ([10] [2] [11]) or 1g-model tests ([12–14]) when function of the friction angle φ. Comparison of the method
the constant λ is set equal to the unity, i.e. as suggested by of slices with experimental data (h >> H, B/H = 1,
Terzaghi and Jelinek [15]. In Fig. 5, the marked rectangles D = 2H/兹苵π). The tests by Berthoz et al. [14] are the only
show the range of experimental values in cohesionless ones performed with a cohesive frictional material (φ =36°,
sand; the value of the friction angle marked on the x-axis c/γ D = 0.07).
Bild 5. Normalisierte Stützdruck s/γ D eines körnigen Mate-
corresponds to a recent result in a cohesive-frictional
rials (c = 0) und kohesiven Materials in Abhängigkeit des
soil (c/γ D = 0.07). The thick solid line was obtained using
Reibungswinkels φ. Vergleich mit der Lamellenmethode mit
the method of slices assuming a uniform support pressure Versuchsergebnissen (h >> H, B/H = 1, D = 2H/兹苵π). Die
and null unsupported span. The results obtained by Tests von Berthoz et al. [14] sind die einzigen, die mit einen
the method of slices with λ = 1.0 agree very well also kohäsiven Material ausgeführt wurden (φ =36°, c/γ D = 0.07)
with numerical results [8], [4]. Therefore, λ will be
taken equal to 1.0 for all calculations of the following
sections.
The paper in hand outlines the computational the bearing capacity of the wedge is calculated by consid-
method (Section 2), investigates the effects of the unsup- ering the equilibrium of an infinitesimal slice (Section
ported span (Section 3) and of the support pressure distri- 2.3). Both, the load of the prism and the bearing capacity
bution (Section 4), proposes a simple design equation for of the wedge, depend on the inclination of the inclined
considering the effect of the unsupported span (Section 3) slip plane. The critical value of the angle ω (Fig. 1), i.e. the
and shows that the error induced by the common approx- value that maximizes support requirements, will be deter-
imation of uniform support pressure is small in the case of mined iteratively.
closed-shield tunnelling but may be significant in the case
of face reinforcement (Section 4). 2.2 The prism loading

2 Computational model Assuming that the ground is homogeneous and obeys


2.1 Outline Mohr-Coulomb failure condition with the cohesion c and
the angle of internal friction φ, the vertical force at the
In the mechanism under consideration (Fig. 1), failure will wedge-prism interface reads as follows:
occur if the load exerted by the prism upon the wedge ex-
ceeds the force which can be sustained by the wedge at its   −λ tanφ 
h h
 Rγ − c −λ tanφ 
upper boundary. At the limit equilibrium, the prism load is Vsilo = max 0, 1 − e R +σ e R

 λ tanφ   surf
equal to the bearing capacity of the wedge. The prism load 
is calculated based upon silo theory (Section 2.2), while B(e + H tan ω ), (4)

42 geotechnik 36 (2013), Heft 1


G. Anagnostou/P. Perazzelli · The stability of a tunnel face with a free span and a non-uniform support

where σsurf denotes the vertical load at soil surface and R The diagram applies to a circular tunnel face of diameter
is equal to the ratio of the volume of the prism to its D and was calculated by means of Eq. (8) considering a
circumferential area: square cross section of equal area (H = B = 0.886 D).
If the maximum unsupported span is very small
B(e + H tan ω ) (emax < 0.5 m), pre-support measures must be applied in
R= . (5)
(
2 B + e + H tan ω ) order to stabilize the tunnel walls (e.g. horizontal grouted
steel pipes in cohesive frictional soils or horizontal
Eq. (4) ensures that the load exerted by the prism will be columns of jet grouting in cohesionless soils). If the pre-
taken equal to zero (rather than becoming negative), if the support is very stiff and extends sufficiently far ahead of
cohesion is higher than a critical value: the tunnel face, the prism load is transferred mainly to the
soil behind the wedge and to the tunnel lining. In this
ccrl = Rγ (if σsurf = 0) (6) case, the load Vsilo would be very small and could be ne-
glected in the face stability analysis. Otherwise a part of
The model of Fig. 1 presupposes that the ground remains this load is transferred to the wedge depending on the stiff-
stable over the unsupported span. This is possible only if ness and bearing capacity of the presupport and of the
the soil exhibits some cohesion. The critical cohesion ccr2 ground ahead of the face [16]. In general, the prism load
can be estimated based upon silo theory by considering a resulting from the Eq. (4) represents an upper bound for
prism having the horizontal cross-section of the unsup- the loading of the wedge. Due to the uncertainties with
ported span: respect to the loading of the wedge in the presence of
presupport, the present paper considers only inherently
e stable unsupported spans (i.e., e < emax).
ccr2 = γ (if σ surf = 0). (7)
(
2 1 + e/B)
2.3 The bearing capacity of the wedge
One can readily verify that this equation agrees qualita-
tively with experience (the longer the unsupported span, Consider the equilibrium of an infinitesimal slice (Fig. 4).
the higher the necessary cohesion will be). For typical val- The slice is acted upon by the following forces: its weight
ues of conventional excavation of a traffic tunnel in soft dG; the “supporting” force V(z) exerted by the underlying
ground (e = 1 m, B = 10 m, γ = 20 kN/m3) the minimum ground; the “loading” force V(z) + dV exerted by the over-
cohesion according to Eq. (7) amounts to about 10 kPa. lying ground; the forces dN and dT at the inclined slip sur-
Reversing Eq. (7) it is possible to derive the maxi- face; the shear force dTs at the two vertical slip surfaces;
mum unsupported span for fixed ground cohesion: and the supporting force dS. Following [8], adjusted for
the linear support pressure distribution of Eq. (3), we ob-
c 1
emax = (if σ surf = 0). (8) tain the equilibrium condition in the sliding direction:
γ 1 c 
 2 − γB 
  dV  B3  z
B − Λ V =  Mc B2c − M γ B3 γ + Ps δs +
dz  H  B
Fig. 6 shows the normalized maximum unsupported span
emax/D as a function of the normalized cohesion c/γD. (
+ Pc B2c + Ps B2 s0 − 0.5 δs , ) (9)

where the coefficients Λ, Mc, Mγ, Pc and Ps have been


introduced by Anagnostou [8] and are given in the Appen-
dix A.
The vertical force V(z) can be obtained by solving
the differential equation (9) for the boundary condition
V(0) = 0. The bearing capacity of the wedge is identical
with the vertical force in z = H. For homogeneous ground,

V(H) = Cs(1)B2s0 + Cc(1)B2c – Cγ(1)B3 γ – Cδs(1)B2δs, (10)

where the coefficients Cs(1), Cc(1) and Cγ(1) are the same
as in Anagnostou [8] (see Appendix A), while the coeffi-
cient Cδs(1) is a new coefficient, which accounts for the
non-uniformity of the support distribution:

Ps  B 
Cδs(1) = − − 0.5 Cs(1). (11)
Λ  ΛH 
Fig. 6. Normalized maximum unsupported span emax/D
as a function of the normalized cohesion c/γ D according to As shown in the Appendix A, the coefficient Cδs (1) is al-
silo theory ways positive. This result in combination with Eq. (10)
Bild 6. Normalisierte, maximale nicht gestützte Abschlags- means that the bearing capacity of the wedge decreases
länge emax/D als Funktion der normalisierten Kohäsion with increasing δs, i.e. if the support pressure at the crown
c/γ D gemäß Silotheorie is higher than the support pressure at the floor. This

geotechnik 36 (2013), Heft 1 43


G. Anagnostou/P. Perazzelli · The stability of a tunnel face with a free span and a non-uniform support

means that the support pressure distribution of Fig. 7a is With the exception of very shallow tunnels and pro-
more favorable than the distribution of Fig. 7b (the aver- vided that the cohesion is lower than Rγ (i.e., that the
age support pressure, i.e. the support pressure so at the prism is unstable without support), both the expression
tunnel axis, being fixed). for the silo loading and the expression for the support
pressure become considerably simpler:
2.4 The support pressure
σ z(H) R c
= f81 −f , (16)
At limit equilibrium the load exerted by the prism is equal γH H 81 γH
to the bearing capacity of the wedge:
where
V(H) = Vsilo. (12)
e
1+
H tan ω
This represents a linear equation for the support pressure f81 = (17)
s0 because V(H) depends linearly on s0 (Eq. (10)). The λ tan φ
solution reads as follows:
and
s0 c σ (H) δs
= f1 − f2 + f3 z + f4 , (13)
Hγ Hγ Hγ Hγ s0 δs c
= f51 − f4 − f52 , (18)
Hγ Hγ γH
where the coefficients f1, f2 and f3 are the same as in
Anagnostou [8] (see Appendix A). The coefficient where

Cδs(1) B 1 B  R B e 
f4 = = − 0.5 − = f4  , λ , φ, ω  (14) f51 = f1 + f3f81 = f51  , , λ , φ, ω  , (19)
Cs(1) ΛH C v (1) − 1 H  H H H 

and accounts for the effect of the support pressure distrib- B e 


ution. It is easy to show that this coefficient is always posi- f52 = f2 + f3f81 = f52  , , λ , φ, ω  . (20)
H H 
tive and, consequently, that the necessary average support
pressure so increases linearly with δs.
According to Eq. (4), the normalized vertical pres- 3 The influence of the unsupported span
sure exerted by the prism is:
According to the mechanism of Fig. 1, the unsupported
σ z(H) 1 Vsilo span increases the prism loading V(H) of the wedge and
= =
γH γH BH tan ω thus also the support pressure required for stability. This
 R c  effect can be clearly recognized in Fig. 8, which shows the
− 
 −λ tan φ  −λ tan φ 
h h
H γH σ R
required support pressure s (assuming uniform distribu-
max 0, 1 − e R  + surf e
 tion) as a function of the wedge angle ω for two values of
 λ tan φ 
 
 γ H 
 
 e   B e h c σ surf 
 1 + H tan ω  = f7  H , H , λ , φ, ω , H , γH , γH  . (15)
 

One can easily verify that with increasing depth of cover h


the exponential term in Eq. (4) decreases rapidly to zero
with the consequence that the silo pressure and the neces-
sary face support pressure become practically indepen-
dent on h.

(a) (b)

Fig. 8. Support pressure s as a function of the angle ω (face


width B = 10 m, face height H = 5 m, depth of cover h >> H,
unit weight γ = 20 kN/m3, friction angle φ = 25°)
Fig. 7. Face support pressure, (a), with δ s < 0 and, (b), Bild 8. Stützdruck s als Funktion vom Winkel ω (Tunnel-
with δ s > 0 breite B = 10 m, Tunnelhöhe H = 5 m, Überdeckung h >> H,
Bild 7. Stützdruck, (a), mit δ s < 0 und, (b), mit δ s > 0 Raumgewicht γ = 20 kN/m3, Reibungswinkel φ = 25°)

44 geotechnik 36 (2013), Heft 1


G. Anagnostou/P. Perazzelli · The stability of a tunnel face with a free span and a non-uniform support

Fig. 9. Support pressure s as a function of the unsupported


span (parameters as in Fig. 8)
Bild 9. Stützdruck s als Funktion der ungestützten Ab-
schlagslänge (Parameter wie in Bild 8)
Fig. 11. Gradient ds/dc as a function of the friction angle
φ for e = 0 – emax (h >> H, B/H = 1, D = 2H/兹苵π)
Bild 11. Gradient ds/dc als Funktion des Reibungswinkels
φ für e σ = 0 – emax (h >> H, B/H = 1, D = 2H/兹苵π)

for zero unsupported span and for the maximum stable


unsupported span emax. The diagram applies to a circular
tunnel face of diameter D, it was calculated by means of
Eq. (18) for a quadratic cross section of equal area (H =
B = 0.886 D) assuming δs = 0 and covers the relevant
parameter range. It should be noted that emax increases
with the cohesion (Eq. (8)) and, consequently, varies along
each line of Fig. 10. The lines start at the same point at
c = 0 because emax = 0 for cohesionless soils.
The linearity of the s(c) relationship was discussed by
Anagnostou [8] for the case of e = 0. The linearity exists al-
so when the unsupported span is set equal to emax (Fig. 10)
or to a fixed fraction of it. Fig. 11 shows the gradient of the
s(c) lines as a function of the friction angle φ for e = 0, 0.5
emax and emax. The gradient ds/dc depends almost linearly
on the ratio e/emax.
A simple formula that approximates with sufficient
Fig. 10. Normalized support pressure s/γ D as a function of accuracy the support pressure s taking into account the ef-
the normalised cohesion c/γ D for e = 0 – emax (h >> H, B/H
fect of the unsupported span is:
= 1, D = 2H/兹苵π)
Bild 10. Normalisierter Stützdruck s/γ D als Funktion der
normalisierten Kohäsion c/γ D für e = 0 – emax (h >> H, B/H (
s = 0.05 cot φ )1.75 γD − cot φ c +
= 1, D = 2H/兹苵π)
 c 
+ ( 0.5cot φ − 0.5)  0.5 − γe 0 ≤ e ≤ emax . (21)
 γD 
the unsupported span. According to Fig. 8, the critical
wedge (i.e. the wedge corresponding to the angle ω that For e = 0 this formula reduces to the equation introduced
maximizes the support pressure) becomes slightly steeper by Anagnostou [8].
in the presence of an unsupported span. Fig. 9 shows the
required support pressure s (i.e., the one corresponding to 4 The influence of the support pressure distribution
the critical wedge assuming uniform distribution) as a 4.1 Support pressure
function of the unsupported span for e < emax. The circles
on every curve mark the maximum stable unsupported For the sake of simplicity and without loss of generality,
span emax (Eq. (8)) and thus the validity range of the pre- we investigated the effect of the support distribution only
sent method (see Section 2.2). for the case of null unsupported span (e = 0). This as-
Fig. 10 shows the normalized uniform support pres- sumption is strictly true only for mechanized tunnelling
sure s/γD as a function of the normalized cohesion c/γD through very weak ground, where the soil closes the gap

geotechnik 36 (2013), Heft 1 45


G. Anagnostou/P. Perazzelli · The stability of a tunnel face with a free span and a non-uniform support

(a) (c)

(b) (d)

Fig. 12. (a) and (c) Normalized support pressure s0/γ D at the tunnel axis and, (b) and (d), support pressure error due to the
assumption of uniform distribution as a function of the normalized gradient of the support distribution δs/γ D for different
values of the shear srength parameters (e = 0, h >> H, B/H = 1, D = 2H/兹苵π)
Bild 12. (a) und (c) normalisierter Stützdruck s0/γ D auf der Tunnelachse und, (b) und (d), Stützdruckfehler infolge der
Annahme einer gleichmässigen Verteilung als Funktion des normalisierten Gradienten der Stützdruckverteilung δs/γ D für
verschiedene Werte der Schubkraftparameter (e = 0, h >> H, B/H = 1, D = 2H/兹苵π)

around the shield and the shield supports the tunnel – The solid straight lines represent the support pressure s0
crown up to the face. that is necessary to stabilize the entire face (zf = 0). They
Fig. 12a shows the normalized support pressure at were calculated by means of Eq. (18) considering a qua-
the tunnel axis s0/γD as a function of the normalized sup- dratic cross section of equal area (H = B = 0.886 D). The
port pressure gradient δs/γD for cohesionless soils and dif- linearity between s0 and δs was observed already in Sec-
ferent values of the friction angle φ. The unit weight γ and tion 2 (see Eqs. (10), (11), (13) and (14)).
the support pressures s0 and δs to be considered in the di- – The dashed curves represent the maximum support
mensionless variables of this diagram depend on whether pressure at the tunnel axis necessary to stabilize smaller
the tunnel is located above or below the water table. In the wedge at the upper part of the face. They were calcu-
first case, γ, s0 and δs represent the dry unit weight γd and lated by means of Eq. (18) considering rectangular cross
the actual support pressures. In the second case, γ, s0 and sections of smaller heights H* (H* < H = B = 0.886 D).
δs represent the submerged unit weight γ ′ and the excess Eq. (18) gives the support pressure in the middle of the
support pressures, i.e. the difference between support rectangular section, i.e. in a distance of (H – H*)/2
pressure and hydrostatic pressure in the ground ahead of above the tunnel axis. The corresponding support pres-
the face. The diagram applies to a cylindrical tunnel of di- sure s0 at the elevation of the tunnel axis (i.e., the pres-
ameter D and considers the range of support pressure dis- sure on the ordinate axis of Fig. 12a) is derived taking
tributions, which are relevant in the practice (see bottom into account the support pressure gradient.
of Fig. 12 and Appendix B).
When the support pressure at the crown is lower The failure of smaller wedges in the upper part of the
than the average pressure (i.e., when δs < 0 as in Fig. 7a), it wedge is relevant only if the support pressure above the
may happen that smaller wedges in the upper part of the tunnel axis is considerably lower than the average pressure
face are more critical than the entire face (for example, the (i.e., if δs is negative and sufficiently high). Otherwise fail-
wedge with foot at zf in the inset of Fig. 13). The investiga- ure of the entire face represents the critical mechanism.
tions of the present section consider also such failure The height H* of the critical wedge decreases with in-
mechanisms. This is why there are two groups of curves in creasing magnitude of δs and depends also on the friction
the left part of Fig. 12a: angle φ (Fig. 13).

46 geotechnik 36 (2013), Heft 1


G. Anagnostou/P. Perazzelli · The stability of a tunnel face with a free span and a non-uniform support

Fig. 13. Critical height as a function of normalized gradient δs/γ D for cohesionless soils (e = 0, h >> H, D = 2H/兹苵π)
Bild 13. Kritische Höhe als Funktion des normalisierten Gradienten δs/γ D für kohäsionslose Böden (e = 0, h >> H,
D = 2H/兹苵π)

The two straight lines starting from the origin of the cohesion does not affect the failure mechanism in
Fig. 12a show the minimum support pressure s0 consid- terms of the height of the critical wedge and, consequent-
ered in the analyses. Consider the case of a negative sup- ly, Fig. 13 applies also to cohesive-frictional soils.
port pressure gradient (Fig. 7a). For given value of the sup- For slurry and EPB shields (left hand side of Fig. 12b
port pressure difference δs, the support pressure s0 at the and 12d), the uniformity assumption overestimates or un-
axis must be ≥ –0.5 δs. Otherwise the support pressure at derestimates the necessary support pressure only slightly
the crown would be negative, which of course is impossi- (the error amounts to maximum ±10 %). For face support
ble. Similarly, s0 must be higher or equal to 0.5 δs in the by bolts, however, the uniformity assumption may under-
case of positive gradient. The functions s0(δs) in Fig. 12a estimate considerably the necessary support pressure.
are, therefore, located above the two limiting lines. Fig. This underlines the importance of taking into account the
12a shows that, for fixed shear strength parameters, the actual support distribution in the case of face reinforce-
support pressure s0 at limit equilibrium is higher than the ment.
minimum support pressure of 0.5 |δs| only within a limited
range of the gradient δs/γD. Outside this range, the mini- 4.3 Special case of support by compressed air
mum support pressure of 0.5 |δs| stabilizes the face with a
safety factor higher than the unity. The case of support by compressed air is insofar special as
just fulfilling the criterion of water pressure compensation
4.2 Error of uniformity assumption at the bottom of the face automatically leads to such a
high excess support pressure above the floor (Fig. 2b), that
The computational method of the present paper allows to face stability is ensured for all practical cases. This can be
quantify the error induced by the usual simplifying as- readily verified from Fig. 12a, which applies to the most
sumption of support pressure uniformity. The error unfavourable case (cohesionless soils). Under the mini-
mum air pressure (which is equal to the hydrostatic pres-

err =
()
s0 0 − s0 δ s( ), sure at the tunnel floor), s0 = 0.5 γw D and δs = γw D. Con-

( )
(22) sequently, the normalized pressures and gradients of
s0 δ s
Fig. 12a amount to 0.5 γw/γ′ and γw/γ′, respectively, and
the (s0/γ′D, δs/γ′D)-points are located on the limiting line
where s0(0) and s0(δs) denote the required support pres- s0 = 0.5 δs and, more specifically, in its rightmost interval,
sure in the case of uniform or linear distribution, respec- where γw/γ′ = 0.8–1. As this interval is above the
tively. s0(δs) accounts also for failure of smaller wedges at lines which fulfill limit equilibrium for the practical
the upper part of the face. A positive value of the error relevant values of the friction angle (φ > 20°), the mini-
means that the uniformity assumption is on the safe side. mum air pressure stabilizes the face with a safety factor
Fig. 12b shows the error as a function of the support Fs > 1.
pressure gradient ds for cohesionless soils with φ = 15–35°. The safety factor Fs is defined here as the factor by
The results will be discussed below, after introducing which the shear strength parameters c and tanφ must be
Fig. 12c and 12d, which apply to the case of cohesive-fric- reduced in order that the considered mechanism is at lim-
tional soils with φ = 25°. The straight lines in Fig. 12c con- it equilibrium [9]. The safety factor can be estimated
cern failure of the entire face and are parallel because the graphically by means of Fig. 14, which shows the relation-
cohesion does not affect the gradient ds0/dδs (see Eqs. ship between the shear strength parameters (tanφ, c/γ′D)
(10), (11), (13) and (14)). It was observed, furthermore, that at limit equilibrium (i.e. for Fs = 1), when the air pressure is

geotechnik 36 (2013), Heft 1 47


G. Anagnostou/P. Perazzelli · The stability of a tunnel face with a free span and a non-uniform support

sumption is, nevertheless, big in the case of face reinforce-


ment.
In the case of compressed air application, the air
pressure needed just to balance the hydrostatic pressure is
sufficient for stabilizing the face with a high safety factor.

Appendix A. Coefficients and derivations

Coefficients in the equilibrium equation


The coefficients appearing in Eq. (9) read as follows:

2 λ tan φ
Λ= , (A1)
cos ω − sin ω tan φ

Λ tan ω
Mc = , (A2)
λ tan φ

Fig. 14. Relationship between the shear strength parameters Mγ = tan ω, (A3)
of the soil (tanφ, c/γ ′D) at limit equilibrium (Fs = 1) assu-
ming that the excess support pressure s0 = 0.5 γwD Λ
Bild 14. Beziehung zwischen den Schubkraftparametern des Pc = , (A4)
2 λ tan φ cos ω
Bodens (tanφ, c/γ ′D) im Grenzgleichgewicht (Fs = 1) unter
der Annahme, dass der Stützdruck-Überschuss s0 = 0.5 γwD
ist Ps = tan (φ + ω). (A5)

Coefficients in the bearing capacity equation


The coefficients Cs(1), Cc(1) and Cγ(1) appearing
equal to the hydrostatic pressure at the tunnel floor. Con- in Eq. (10) read as follows:
sider, for example, a soil with submerged unit weight γ′ =
γw and the shear strength parameters given by the point A C v (1) − 1
Cs (1) = Ps (A6)
in Fig. 14. As the point A is above the curve for γw/γ′ = 1, Λ
the safety factor Fs of the tunnel face is higher than one.
Reducing the strength parameters by the same factor, the F(1)
C γ (1) = Mγ (A7)
point A moves along the line OA. The safety factor Fs is Λ2
the reduction factor which corresponds to point B and
can be determined easily as indicated in Fig. 14. C v (1) − 1 F(1)
Cc (1) = Pc + Mc (A8)
Λ Λ2
5 Conclusions
where
The 3D failure mechanism under consideration allows to
take into account the possibility of an unsupported span Cv(1) = e(ΛH/B), (A9)
between lining and face and a linear distribution of the
face support pressure over the height of the face. ΛH
F(1) = C v (1) − 1 − . (A10)
The unsupported span increases the vertical load on B
the soil ahead the face and consequently also the support
pressure required to stabilize the face. This effect can be Sign of coefficient Cδs (1)
significant also for the relatively narrow unsupported This coefficient is given by Eq. (11) as a function of
spans (e = 0.5–1.5 m) of conventional soft ground tun- the coefficients Ps, Cs(1) and Λ. Taking into account
nelling. A new term was added in the design equation of Eqs. (A6) and (A9),
Anagnostou [8] for considering the contribution of the un-
Ps  et 2
supported span.
When the face support pressure decreases from the
()
Cδs 1 = 1 − 2
2Λ  t
+ et + 
t
(A11)
tunnel floor to the tunnel crown, as in slurry- or EPB-
shield tunnelling, the failure of a smaller part of the face where
adjacent to the crown may be more critical than the fail-
ure of the entire face. The height of the critical wedge de- ΛH
t= . (A12)
pends on the gradient of the support pressure and on the B
friction angle of the soil.
The usual simplifying assumption of uniform sup- For the relevant range of the failure angle ω (0 < ω < π/2 –
port pressure is reasonable for the typical support pres- φ), the coefficient Λ is always positive and, consequently,
sure gradients of slurry- or EPB-shield tunnelling, because the ratio Ps/2Λ and the variable t are always positive too.
it leads only to a slight over- or underestimation of the sup- By replacing the exponential function et by a Taylor series
port pressure. The error of the simplifying uniformity as- and taking into account that t > 0, Eq. (A11) becomes:

48 geotechnik 36 (2013), Heft 1


G. Anagnostou/P. Perazzelli · The stability of a tunnel face with a free span and a non-uniform support

δs γ − γw
 2
1 −  1 + +
t t2
+ ... +
tn  
+ ... +  γD
≅− s
γ′
≅ − 0.1 − 0.3 ( ) (B4)
P  t 1! 2! n!  
()
Cδs 1 = s 
2 Λ   2
=
for γw = 10 kN/m3 and γ ′ = 6–13 kN/m3.
t t2 tn 
 1 + + + ... + + ... + 
 1! 2! n!  t 
Support by EPB
Ρ  2 t 2
 2 t 3
 2 t n  Above the water table (Fig. 2e) the normalized support
= s  1 −  +  1 −  ... +  1 − + ...  (A13)
2Λ  3  2!  4  3!  n + 1  n!  pressure gradient

δs γ
≅ − m ≅ − (0.7 − 1), (B5)
As Ps/2Λ as well as the terms within the brackets are posi- γD γd
tive, the coefficient Cδs (1) is positive too.
depending on the degree of loosening or compaction of
Coefficients of the support pressure equation the excavated soil in the working chamber. Below the
The coefficients appearing in Eq. (13) read as follows: water table (Fig. 2f),

B  δs γ − γw
B C γ (1) ≅− m ≅ − (0.7 − 1). (B6)
f1 = = f1  , λ , φ, ω  , (A14) γD γ′
H Cs (1) H 

Notation
Cc (1) B 
f2 = =f , λ , φ, ω  , (A15)
Cs (1) 2  H  B Width of the rectangular tunnel face
c Cohesion of the ground
ccr1 Critical cohesion in order that the prism remains
tan ω B 
f3 = = f3  , λ , φ, ω  . (A16) stable
B H  ccr2 Critical cohesion in order that the unsupported
C (1)
H s span remains stable
Cs Coefficient (Eq. A6)
Appendix B. Range of normalized support pressure gradient Cc Coefficient (Eq. A8)
Cv Coefficient (Eq. A9)
Support by face reinforcement above the water table Cγ Coefficient (Eq. A7)
Taking account of Eq. (1) with H = D, the normalized sup- Cδs Coefficient (Eq. 11)
port pressure gradient e Unsupported span
emax Maximum stable unsupported span
δs n H tan ω πdτ m n tan ω πdτ m d Grouted borehole diameter
= ≅ . (B1)
γD γd D γd f1 Coefficient (Eq. A14)
f2 Coefficient (Eq. A15)
For ω = 30°, n = 0.5 bolts/m2, d = 0.10 m, τm = 100 kPa f3 Coefficient (Eq. A16)
and γd = 20 kN/m3, we obtain δs/γD = 0.45. Considerably f4 Coefficient (Eq. 14)
higher or lower values are possible depending on the angle f51 Coefficient (Eq. 19)
ω, on the bond strength τm and on the reinforcement den- f52 Coefficient (Eq. 20)
sity n. f7 Coefficient (Eq. 15)
f81 Coefficient (Eq. 17)
Support by compressed air F Coefficient (Eq. A10)
Above the water table, the support pressure is uniform Fs Safety factor of the tunnel face
(Fig. 2a) and, consequently, δs/γD = 0. Below the water G Weight
table, the excess support pressure increases linearly with h Depth of cover
the elevation above the floor (Fig. 2b). The normalized H Height of the rectangular tunnel face
support pressure gradient H* Height of a smaller wedge at the upper part of the
face
δs γ w D γ w L′ Bolt length (Fig. 3)
= = ≅ 0.8 − 1. (B2)
γD γ ′ D γ′ Li′ Anchorage length inside the wedge (Fig. 3)
Lo′ Anchorage length outside the wedge (Fig. 3)
Support by pressurized slurry n Density of the face reinforcement
Above the water table (Fig. 2c) the normalized support N Normal force
pressure gradient R Ratio of the volume to circumferential area of the
prism
δs γ
γD γd
(
≅ − s ≅ − 0.5 − 0.8 ) (B3) s
S
Support pressure
Support force
s0 Support pressure at the tunnel axis
for γs = 11–12 kN/m3 and γd = 14–21 kN/m3. Below the T Shear force at the inclined slip plane
water table (Fig. 2d), Ts Shear force at the lateral slip plane

geotechnik 36 (2013), Heft 1 49


G. Anagnostou/P. Perazzelli · The stability of a tunnel face with a free span and a non-uniform support

V Vertical force [5] Cornejo, L.: Instability at the face: its repercussions for tun-
Vsilo Vertical load exerted by the prism upon the wedge nelling technology. Tunnels and Tunnelling, April (1989), No.
x Horizontal co-ordinate paralelly to tunnel axis 21, pp. 69−74.
[6] Vermeer, P. A., Ruse, N.: Die Stabilität der Tunnelortsbrust in
y Horizontal co-ordinate perpendicularly to tunnel
homogenem Baugrund. Geotechnik 24 (2001), No. 3, pp. 186−
axis
193.
z Vertical co-ordinate [7] Li, Y., Emeriault, F., Kastner, R., Zhang, Z. X.: Stability analy-
zf Elevation of the wedge foot sis of large slurry shield-driven tunnel in soft clay. Tunnelling
and Underground Space Technology 24 (2009), No. 4, pp.
Greek symbols 472−481.
γ Unit weight of the soil [8] Anagnostou, G.: The contribution of horizontal arching to
γ′ Submerged unit weight of the soil tunnel face stability. Geotechnik 35 (2012), No. 1, pp. 34–44.
γd Dry unit weight of the soil [9] Anagnostou, G., Kovári, K.: The face stability of slurry-shield
γm Unit weight of the excavated material in the work- driven tunnels. Tunnelling and Underground Space Technolo-
ing chamber gy 9 (1994), No. 2, pp. 165−174.
[10] Chambon, P., Corte, J. F.: Shallow tunnels in cohesionless
γs Unit weight of the slurry
soil: stability of tunnel face. ASCE J. Geotech. Eng. 120 (1994),
γw Unit weight of the water
No. 7, pp. 1148−1165.
δs Difference between support pressure at the crown [11] Plekkenpol, J. W., van der Schrier, J. S., Hergarden, H. J.:
and at the floor Shield tunnelling in saturated sand-face support pressure and
λ Coefficient of lateral stress soil deformations. In: van Lottum, H., Bezuijen, A. (eds.): Tun-
Λ Coefficient (Eq. A1) nelling: a decade of progress. London: Taylor & Francis, 2006.
Μc Coefficient (Eq. A2) [12] Kirsch, A.: Experimental investigation of the face stability of
Μγ Coefficient (Eq. A3) shallow tunnels in sand. Acta Geotechnica 5 (2010), No. 1, pp.
Ρc Coefficient (Eq. A4) 43–62.
Ρs Coefficient (Eq. A5) [13] Messerli, J., Pimentel, E., Anagnostou, G.: Experimental stu-
dy into tunnel face collapse in sand. In: Springman, S., Laue, J.,
σsurf Soil surface load
Seward. J. (eds.) Physical Modelling in Geotechnics, Vol. 1, Zü-
σy Horizontal stress perpendicularly to the tunnel axis
rich, 2010, pp. 575−580.
σz Vertical stress [14] Berthoz, N., Branque, D., Subrin, D., Wong, H., Humbert, E.:
τm Bond strength of the soil – grout interface Face failure in homogeneous and stratified soft ground: Theo-
φ Friction angle of the ground retical and experimental approaches on 1g EPBS reduced sca-
ω Angle between face and inclined sliding plane of le model. Tunnelling and Underground Space Technology 30
the wedge (2012), pp. 25−37.
ωcrit Angle between face and inclined sliding plane of [15] Terzaghi, K., Jelinek, R.: Theoretische Bodenmechanik. Ber-
the critical wedge lin: Springer-Verlag, 1954.
[16] Anagnostou, G.: Standsicherheit im Ortsbrustbereich beim
References Vortrieb von oberflächennahen Tunneln. Symposium „Städti-
scher Tunnelbau – Bautechnik und funktionelle Ausschrei-
[1] Anagnostou, G., Serafeimidis, K.: The dimensioning of tun- bung“, S. 85−95, ETH Zürich, 1999.
nel face reinforcement. In: Barták, J., Hrdina, I., Romancov,
G., Zlámal, J. (eds.): Underground Space – the 4th Dimension Authors
Prof. Dr. Georgios Anagnostou
of Metropolises. ITA World Tunnel Congress, Prague, 2007, pp.
ETH Zürich
291−296. Institute for Geotechnical Engineering
[2] Idinger, G., Aklik, P., Wu, W., Borja, R. I.: Centrifuge model Wolfgang-Pauli-Strasse 15
test on the face stability of shallow tunnel. Acta Geotechnica 6 CH-Zürich 8093
(2011), No. 2, pp. 105–117.
[3] Mollon, G., Dias, D., Soubra, A.-H.: Rotational failure mecha- Dr. Paolo Perazzelli
nisms for the face stability analysis of tunnels driven by a pres- ETH Zürich
surized shield. Int. J. Numer. Anal. Meth. Geomech. 35 (2011), Institute for Geotechnical Engineering
No. 12, pp. 1363–1388. Wolfgang-Pauli-Strasse 15
[4] Perazzelli, P., Anagnostou, G.: Comparing the limit equilibri- CH-Zürich 8093
um method and the numerical stress analysis method of tunnel
face stability assessment. In: 7th Int. Symp. on „Geotechnical Submitted for review: 7 September 2012
Aspects of Underground Construction in Soft Ground“. Rome, Revised: 30. November 2012
2011. Accepted for publication: 30. November 2012

50 geotechnik 36 (2013), Heft 1

Das könnte Ihnen auch gefallen