Professional Documents
Culture Documents
International Journal of
Rock Mechanics & Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms
art ic l e i nf o a b s t r a c t
Article history: In the vicinity of a tunnel face, the fictitious support pressure represents the support pressure acting on
Received 6 February 2015 the periphery of the tunnel provided by the rock mass itself. It is an indicator of the self-supporting
Received in revised form capacity of the rock mass.This paper aims at solving the fictitious support pressure on a circular tunnel
25 June 2015
with hydrostatic initial stress field. A numerical procedure for the fictitious support pressure is proposed
Accepted 3 July 2015
for the elastic-perfectly-plastic, elastic-brittle, and strain-softening rock masses. The procedure is com-
posed of two steps: first, the ground reaction curve (GRC) and the longitudinal deformation profile (LDP)
Keywords: are solved by a modified numerical approach; in this step, the finite difference method is utilised to
Fictitious support pressure derive the strain components, the stress components, and the radius of the plastic zone; then, by cou-
Circular opening
pling the GRC and LDP, a fictitious support pressure is obtained by a simplified approach. By using the
Tunnel face effect
proposed procedure, the influencing factors of the fictitious support pressure, such as the critical plastic
Finite difference method
Strain-softening behaviour. softening parameter, the rock mass quality, the dilatancy angle and the initial stress condition are studied
individually. The results indicate that the effect of the critical plastic softening parameter on the fictitious
support pressure, especially for the rock mass with good quality is obvious. For a rock mass with strong
dilatancy behaviour, the fictitious support pressure ahead of the tunnel face decreases rapidly. The elastic
assumption of the rock mass behaviour will underestimate the stress relief factor by a large extent.
& 2015 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijrmms.2015.07.001
1365-1609/& 2015 Elsevier Ltd. All rights reserved.
L. Cui et al. / International Journal of Rock Mechanics & Mining Sciences 78 (2015) 336–349 337
the final stage, pf decreases to zero and the interaction process distributed along the excavation boundary. Before the support
ceases. installation, pi is equal to pf. During the interaction process of the
The pf along the tunnel axis represents the tunnel face effect ground-support system, pi is composed of the fictitious support
from a quantitative standpoint. The schematic diagram to solve pf pressure pf and the support pressure ps, which are provided by the
is presented in Fig. 2. As shown in Fig. 2, u0 at a given distance x of tunnel face effect and the installed support, respectively.
the LDP corresponds to the pf in the GRC. Hence, prior to pre-
dicting pf, the LDP and GRC should be presented first. 2.3. Failure criterion and the flow rule
Fig. 3. Distribution of plastic zone and corresponding stress–strain relationship (a) for EPP rock mass (b) for SS rock mass (c) for EB rock mass.
in which sci is the uniaxial compression strength of the rock mass where η* is the critical plastic softening parameter. The paramter
in the intact state; mb, s and a are the strength parameters. η* is infinitely large for the EPP rock mass, and it is equal to 0 for
The H–B and M–C failure criteria for EPP, SS, EB rock masses can the EB rock mass. The parameters ωpeak and ωres are the peak and
be rewritten as residual values of a given strength parameter. For the SS rock mass,
ω reduces linearly with the increase in η during the plastic soft-
a(η)
f (σθ, σr, η) = σθ − σr − σci⎡⎣mb(η)σr/σci + s(η)⎤⎦ ening stage, and remain constant beyond the critical value η* (i.e.
in the plastic residual stage). For the EPP and EB rock masses, ω is
(H−B failure criterion) (6a) constant in the plastic or plastic residual zone.
ω (η) = ω peak (EPP rock mass) (7a) where Kψ is the dilatancy coefficient, i.e.
1 + sin ψ
⎧ peak η Kψ =
⎪ω − (ω peak − ωres) , 0< η < η* 1 − sin ψ (9)
ω(η) = ⎨ η* (SS rock mass)
⎪ ωres, η ≥ η*
⎩ (7b) and ψ is the dilatancy angle.
340 L. Cui et al. / International Journal of Rock Mechanics & Mining Sciences 78 (2015) 336–349
3. Determination of GRC 3.2.2. Stress and strain components in the elastic and plastic zones
In the elastic zone, the closed-form solutions of the stress
The closed-form solution of the GRC for EPP and EB rock components and strain components have been proposed by Ref. 25
masses can be obtained because the strength parameters in the According to Hooke's law, the elastic strain increments in the
plastic (or plastic residual) zone are constant. Relevant solutions plastic zone are:
can be found in Refs. 7,9,24-26,35. The derivation of the GRC for ⎧ e ⎫
⎪ Δε r (i) ⎪ 1 + μ ⎡1 − μ − μ ⎤ ⎧ Δσr (i) ⎫ ⎪ ⎪
the SS rock mass seems to be more complicated, which should be ⎨ ⎬= ⎢ ⎥⎨ ⎬
solved by numerical methods8,10-12,25. For instance, Lee and
⎪ e ⎪
⎩ Δε θ(i) ⎭ E ⎣ − μ 1 − μ ⎦ ⎩ Δσθ(i) ⎭ ⎪ ⎪
(14)
Pietruszczak12 divided the potential plastic zone into a finite
in which E and μ are the Young's modulus and Poisson's ratio of
number of concentric rings, and calculated the increments of
the rock mass; Δεθ(e i) and Δεre(i) are the tangential and radial elastic
stress and strain for each ring in a successive manner. Carranza–
strain increments at the ith annulus.
Torres8 and Alonso et al.10 analysed the strain-softening rock mass
The increment of plastic softening parameter Δη(i) at the i th
by the self-similarity method. In this section, a modified numerical
annulus can be described as (referring to Eq. (3))
approach is proposed to solve the GRC in a simpler way.
Δη(i) = Δε θ(pi) − Δε rp(i) (15)
3.1. Finite difference method where Δεθ(p i) and Δεrp(i) are the tangential and radial plastic strain
increment at the ith annulus.
The finite difference method (FDM), proposed in Refs. 12 and In accordance with the non-associated flow rule, the relation
25 is utilised to cope with this issue. The plastic zone (also in- between Δεrp(i) and Δεθ(p i) is
cluding the plastic softening zone and plastic residual zone) is
Δε rp(i) = − K ψ Δε θ(pi) (16)
divided into a set of concentric annuli, where r(i) and r(i-1) are the
radii of the inner and outer boundaries of the ith annulus. At the In the plastic zone, the strain increment includes the plastic
outer boundary of the plastic zone, sr(0) and sθ(0) are equal to sr2 strain increment and elastic strain increment, then, the equation
and sθ2 at the elasto-plastic boundary. A constant radial stress can be written as
increment Δσr is assumed for each annulus, i.e.
ε r (i) + K ψ ε θ(i) = ε r (i − 1) + K ψ ε θ(i − 1) + Δε re(i) + K ψ Δε θ(e i) + Δε rp(i)
σr2 − pi
Δσr (i) = σr (i − 1) − σr (i) =
n (10) + K ψ Δε θ(pi) (17)
where n is the total number of the concentric annuli; and sr(i) In terms of the small strain case, the displacement compat-
denotes the radial stress at r = r(i) (i ¼ 0, 1, …, n). ibility is
du u
εr = , εθ =
3.2. Modified numerical approach dr r (18)
(H−B failure criterion) (11) where u(i) is the radial displacement at r ¼ r(i).
By combining Eqs. (14), (17) and (19), the radial displacement
(
σr2 = 2σ0 − 2C peak Kφ peak / Kφ peak + 1 )( ) u(i) can be expressed as:
Table 1 Table 2
Parameters of rock masses for verification of LDP. Parameters of rock masses for verification of GRC
GSImax 75 60 50 40 40 GSImax – – – –
GSImin 40 35 30 27 – GSImin – – – –
mp 2.87 1.68 1.17 0.821 0.821 mp 7.5 1.7 2 0.5
sp /10 3 62.2 11.7 3.9 1.3 1.3 sp /10-3 100 3.9 4 1.0
mr 0.821 0.687 0.575 0.516 – mr 1 1 0.6 0.1
sr /10 3 1.3 0.7 0.4 0.3 – sr /10-3 10 0 2 0.5
φp /deg 29.52 25.68 23.13 20.64 20.64 φp /deg – – – –
C p / MPa 3.637 2.673 2.242 1.878 1.878 C p /MPa – – – –
φr /deg 20.64 19.42 18.21 17.49 – φr /deg – – – –
C r / MPa 1.878 1.707 1.536 1.432 – C r /MPa – – – –
ψ /deg 7.38 4.49 2.89 1.55 – ψ /deg – – – –
Kψp – – – – – Kψp – – 1.698 1.698
Fig. 5. Comparison of radial displacement (a) u0 E /R0 σ0 versus r /R0 with case A2 (b) u0 E /R0 σ0 versus r /R0 with case B2 (c) GRC with case C2 (d) GRC with case D2.
for the GRC is practicable for the three types of rock mass. solution3 by H–B failure criterion. Compared with the analytical
solution for the M–C failure criterion, the proposed solution is
6.1.2. LDP closer to the FLAC2D solution3.
In Ref. 3, numerical and analytical approaches were proposed In order to show the advantage of the proposed approach over
to obtain the LDP. The numerical approach was conducted by use the analytical approach3, Fig. 7 plots the values of Rpmax for the SS
of FLAC2D code with Neumann boundary condition (applied rock mass by several approaches. The proposed approach, the
stress) and FLAC3D code with Dirichlet boundary condition (fixed numerical approach (FLAC2D and 3D) and the analytical approach3
displacement). For the analytical approach, Rpmax of the EPP rock are involved. The self-similar approach for M–C failure criterion
mass was obtained according to the solution in Ref. 8. Rpmax of the from Ref. 10 is included to make comparison. Fig. 7 shows that the
proposed solution tends to be closer to the numerical solutions by
SS rock mass was estimated according to the minimum square
Refs. 3 and 10. With low values of GSI, the analytical solution
fitting processes of statistical study for 400 tunnels. Then, the LDP
overestimates Rpmax to some extent. On the whole, in contrast to
of the EPP and SS rock masses can be solved by substituting Rpmax
the analytical approach, the results by the proposed approach fit
into Eq. (25). The result of the analytical approach shows good
better with the numerical results by FLAC2D and 3D. This means
agreement with that of the numerical approach (FLAC2D and 3D
the accuracy of the proposed approach for the LDP is acceptable.
codes3). In order to validate the accuracy of the proposed ap-
proach, on the basis of cases D1-E1, the calculated LDPs are com- 6.2. Verification of the fictitious support pressure solution
pared with those obtained by the numerical and analytical solu-
tions in Ref. 3. In fact, the simplified solution of pf has been deduced from a
Fig. 6 shows the comparison of the LDP solution of EPP and SS theoretical standpoint. The objective for verifying the simplified
rock masses by the proposed approach, and the approach taken in solution is to check whether the stress and the strain components
Ref. 3, with H–B and M–C failure criteria, respectively. As shown in refer to the one-to-one correspondence for a given condition. If
Fig. 6, for the EPP rock mass, the results by the proposed approach the stress and strain components remain constant when Δsr is
are highly consistent with those by FLAC2D and 3D codes3. For the very small, the simplified solution will be verified. Therefore, in
SS rock mass, the proposed LDP solution fits well with the FLAC3D this section, the solution of pf is validated by discussing the
344 L. Cui et al. / International Journal of Rock Mechanics & Mining Sciences 78 (2015) 336–349
Fig. 6. Comparison of LDP solution (a) case E1 for EPP rock mass with H–B failure criterion (b) case E1 for EPP rock mass with M–C failure criterion (c) case D1 for SS rock
mass with H–B failure criterion (d) case D1 for SS rock mass with M–C failure criterion.
Fig. 7. Comparison of Rpmax solution for SS rock mass (a) M–C failure criterion (b) H–B failure criterion.
accuracy of the GRC solution. calculated by the strength constants mb, s, and a of H–B failure
Here, cases A3 and B3 with different qualities of the rock mass criterion according to the method introduced in Ref. 23.
are considered. The parameters are listed in Table 3. The equiva- Based on cases A3 and B3, η* is regarded as infinitely large, 0.01,
lent friction angle φ and cohesion C of M–C failure criterion are 0 to represent the EPP, SS and EB rock masses, respectively. The
L. Cui et al. / International Journal of Rock Mechanics & Mining Sciences 78 (2015) 336–349 345
Table 3 Table 5
Parameters of rock masses for cases A3 and B3
Different values of η* for case A3.
A3 B3
A3 ① ② ③ ④ ⑤ ⑥ ⑦ ⑧
GSImax 75 50
ηn EPP 5 0.5 0.05 0.005 0.003 0.002 EB
GSImin 27 26
mi 16.30 12.00
mp 6.044 1.650
sp /10 3 50.987 2.5996
Table 6
ap 0.501 0.506
Different values of η* for case B3.
mr 0.916 0.626
sr /10 3 0.178 0.142
B3 ① ② ③ ④ ⑤ ⑥ ⑦ ⑧
ar 0.527 0.530
φp /deg 45.359 32.4739 ηn EPP 5 0.5 0.05 0.03 0.01 0.001 EB
C p / MPa 6.216 3.077
φr /deg 29.827 24.405
C r /MPa 2.489 1.896
(number of the annuli) for EPP, SS and EB rock masses are listed in
E/GPa 30.0 9.0
μ 0.25 0.25
Table 4; n ranges from 25 to 5000.
R0 /m 5.0 5.0 As displayed in Table 4, the decreasing rate of u0max is re-
σ ci /MPa 110 80 markably reduced as n increases. When n is larger than 3000, the
σ0 /MPa 35 35 values of u0max are basically constant in most cases. Some cases will
not converge to a certain value when n reaches to 5000, whereas it
is acceptable since the decreasing rates of these cases are very
Table 4 small. In a word, the values in Table 4 validate the accuracy of the
Values of u0max with different n: (a) for EPP rock mass (b) for SS rock mass (c) for EB simplified approach for pf.
rock mass
Fig. 8. Variation law of pf versus X * with case A3 (a) X * ranges from 10 to 10 (b) enlarged view of (a).
Fig. 9. Variation law of pf versus X * with case B3 (a) X * ranges from 10 to 10; (b) enlarged view of (a).
Table 7 smaller than that for the elastic condition, and a higher η* gives
Percentage of pf/s0 at X * = 0 for cases A3 and B3. rise to a larger value of pf. Consequently, for a given u0 /u0max (in the
practical tunnel engineering, u0 /u0max is commonly estimated by
① ② ③ ④ ⑤ ⑥ ⑦ ⑧
pf/s0 at X * = 0 (%) the field test data), the stress relieve factor fs will be under-
Fig. 10. Relationship between pf /σ0 and u0 /u0max (a) case A3 (b) case B3.
Fig. 11. Variation law of pf with X * for case B3 (a) X * ranges from 10 to 10 (b) enlarged view of (a).
Table 8 mass give rise to relatively small decreasing rate of pf. The de-
X1* , X2* and X3* for different σ0 . formation of high s0 and weak rock mass near the tunnel face
increases significantly as pf is small. As a result, in order to prevent
s0/MPa 5 20 35 50 65
the rock mass from squeezing the support, the stress near the
X1* 5 6 6 7 8 tunnel face can be released prior to the interaction of the rock
X2* 0.10 0.16 0.98 1.54 1.96 mass and support. This conclusion can be validated by many case
X3* 0.42 0.50 1.21 1.71 histories of tunnels and mines37-40. In these cases, while con-
fronted with the squeezing problems for tunnels excavated in the
soft rock with the high initial stress condition, the workers install
Table 9
flexible or yielding support, or allow the deformation to relieve the
Five values of ψ and Kψ.
high stress. The purpose is to avoid the support buckling or
0 φ/8 φ/4 φ/2 φ breaking down when suffering from the heavy load and large
deformation.
ψ 0 3.56 7.11 14.23 28.45
Kψ 1 1.13 1.28 1.65 2.82
Fig. 12. Variation law of pf with X * for different ψ (a) X * ranges from 10 to 10 (b) enlarged view of (a).
Table 10 Conclusions
X1* , X2* and X3* for different ψ.
In this paper, for the circular tunnel subjected to a hydrostatic
ψ 0 φ/8 φ/4 φ/2 φ
condition, a numerical procedure to solve the fictitious support
6 6 6 6 5 pressure is presented. The procedure is mainly composed of two
X1*
0.84 0.98 1.15 1.55 2.83
parts: a new numerical approach to solve the GRC and LDP, and a
X2*
0.31 0.50 0.71 1.21 2.68
simplified approach to solve the fictitious support pressure. Then,
X3*
by comparing the calculated results with those in other studies, it
is indicated that the proposed procedure is capable of providing
reasonable estimation. In the end, a series of parametric studies is
7.3. Influence of dilatancy coefficient Kψ
carried out and the following conclusions are drawn:
The influence of the critical plastic softening parameter on the
Five values of the dilatancy angle ψ are selected to represent
fictitious support pressure, especially for the rock mass with good
different dilatancy behaviours of the rock mass. The case B3 is
quality, is obvious. The influence of the dilatancy behaviour on the
analysed. It is postulated that ψ is equal to 0, φ/8, φ/4, φ/2, φ,
fictitious support pressure is trivial.
respectively. For the sake of simplicity, the friction angle φ is The elastic assumption of the rock mass behaviour will un-
treated as the average value of the peak friction angle φpeak and derestimate the stress relieve factor during tunnelling. Thus, the
the residual friction angle φres in the case B3; η* is regarded as support design with elastic conditions tends to become unsafe. In
0.01, respectively. The calculated ψ and Kψ are listed in Table 9. addition, for different high initial stress conditions, the stress re-
lieve factors are significant, and basically identical.
7.3.1. Variation law of pf versus X* While the weak and soft rock mass is excavated in the high
initial stress field, the fictitious support pressure and its decreasing
Fig. 12 plots the variation law of pf versus X * for different ψ. It is
rate along the axial direction is relatively small. In order to prevent
observed in Fig. 12 that pf decreases as ψ increases. Fig. 12 reveals
the rock mass from squeezing the support, the high stress near the
that pf decreases rapidly far away ahead of the tunnel face when ψ
tunnel face can be appropriately released prior to the interaction
reaches φ. The area will become unstable due to the remarkable
of ground-support system.
reduction of pf. For the rock mass with high initial stress condition and strong
dilatancy behaviour, the plastic softening and residual zones
7.3.2. X1*, X2*, and X3* emerge far ahead of the tunnel face. In this case, the pre-re-
The values of X1*, X2* and X3* for different ψ are displayed in inforcement should be installed, allowing the rock mass to deform
Table 10. It is found that X1* remains constant until ψ increases to in a controlled manner and mobilise its strength.
φ. Consequently, ψ tends to have negligible impact on the duration
of the tunnel face effect. In contrast, the influence of ψ on X2* and
Acknowledgement
X3* are more obvious. For instance, the absolute values of X2* and X3*
for ψ = φ are 2.83 and 2.68, which are 3.37 and 8.65 times as large
The authors would like to express their gratitude to the Na-
as those for ψ = 0, respectively. As listed in Table 8, the absolute
tional Natural Science Foundation of China (51278216 and
values of X2* and X3* for s0 ¼ 60 MPa are 1.98 and 1.71, respectively.
51478201) for the financial support, and to Professor Phil Dight for
This indicates that for the rock mass with high initial stress con-
providing assistance with English usage.
dition, or strong dilatancy behaviour, the plastic softening and
residual zones emerge far ahead of the tunnel face. In this case, the
rock mass in vicinity of tunnel tends to be unstable. In an effort to
References
prevent the rock mass from collapse, the pre-reinforcement
should be installed. This allows the rock mass to deform in a 1 Carranze-Torres C, Fairhurst C. Application of the convergence-confinement
controlled manner and mobilise its strength. method of tunnel design to rock masses that satisfy the Hoek–Brown failure
L. Cui et al. / International Journal of Rock Mechanics & Mining Sciences 78 (2015) 336–349 349