Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Tunnel-Pile Interaction Analysis Using Cavity

Expansion Methods
Alec M. Marshall1

Abstract: Evaluation of the impact of tunnel construction on existing buried structures is an important problem. This paper presents an an-
alytical method for estimating the effect of tunnel construction on end-bearing piles located above the tunnel. The method can be used to es-
timate the safe relative distance between existing piles and newly constructed tunnels. Spherical and cylindrical cavity expansion/contraction
analyses are used to evaluate pile end-bearing capacity and the reduction of confining pressure at the pile tip that results from tunnel volume loss.
Pile end-bearing capacity is then reevaluated based on the reduced confining pressure at the pile tip. A modified shear modulus is used to account
for the effect of pile installation on soil stiffness. The method is used to analyze centrifuge experiments conducted to study this problem. For
the experimental data, where the service load applied to the piles during tunnel volume loss ranged between 50 and 60% of the maximum jack-
ing force, the analytical method showed that pile failure occurred when the load-carrying capacity was reduced below 80% of its original value.
A parametric analysis is included that highlights the effect of key soil properties on results. DOI: 10.1061/(ASCE)GT.1943-5606.0000709.
© 2012 American Society of Civil Engineers.
CE Database subject headings: Tunnels; Piles; Cavitation; Soil-structure interactions; Centrifuge.
Author keywords: Tunnel; Pile; Cavity expansion; Soil-structure interaction; Centrifuge.

Introduction assumption that greenfield displacements can be used as the input for
the evaluation of the soil-structure interaction problem is ques-
The construction of tunnels within congested urban areas inevitably tionable. For less intrusive pile installation methods like bored piles,
affects existing surface and subsurface structures and infrastructure. it may be that this assumption holds. For jacked or driven piles,
Determination of these effects is difficult to accurately assess be- however, Marshall and Mair (2011) showed that greenfield con-
cause of the nonlinear behavior of soil and the complex nature of ditions are significantly altered by the pile installations. They
the soil-structure interactions that take place. Analytical solutions concluded that greenfield displacements should not be used as an
for these types of problems have the benefit of computational effi- input for tunnel-soil-pile interaction analyses relating to jacked or
ciency over more robust numerical methods; however, fundamental driven piles.
simplifying assumptions of material behavior (e.g., elasticity) and The problem of tunnel construction beneath piles is becoming
soil-structure interactions are generally made to obtain solutions. increasingly important in urban areas where available underground
Despite this, many analytical solutions have proven to be useful. For space is limited. In addition, because the overall cost of constructing
example, the Winkler foundation models presented in Attewell et al. deep transport tunnels in urban areas is generally greater than for
(1986) are often used to evaluate the effect of tunneling on buried shallow tunnels, a tunnel route that lies closer to building founda-
pipes, and the elastic continuum solutions of Klar et al. (2005) have tions is generally financially preferable. The important consideration
been validated using centrifuge experiment data (Vorster et al. is how close a tunnel can be constructed to an existing pile. The
2005), as well as numerical modeling (Klar and Marshall 2008). answer to this question involves understanding the load-carrying
Attewell et al. (1986) also used a Winkler model to evaluate the capacity of the pile and how the relaxation of stresses within the
effect of tunneling on buildings, and Potts and Addenbrooke (1997) ground that result from tunnel volume loss will affect the pile
and Franzius et al. (2006) evaluated tunneling-induced damage to capacity. The tunnel-soil-pile interaction problem has been anal-
buildings using a relative stiffness approach. These solutions all yzed in a variety of ways: experimentally (Morton and King 1979;
use greenfield tunneling displacements (i.e., displacements caused Bezuijen and Van der Schrier 1994; Jacobsz 2002; Meguid and
by tunneling in soil with no other structures affecting results) as an Mattar 2009; Marshall 2009; Marshall and Mair 2011), analytically
input and evaluate the change in greenfield displacements due to the (Chen et al. 1999; Poulos and Deng 2004; Kitiyodom et al. 2005),
existence of a structure (i.e., a pipeline or building). For the problem and numerically (Cheng et al. 2007; Lee and Jacobsz 2006; Mroueh
of tunneling beneath piles, which is the focus of this paper, the and Shahrour 2002).
Analytical solutions offer an attractive means of studying the
1
problem because of their computational efficiency. Chen et al.
Lecturer, Dept. of Civil Engineering, Faculty of Engineering, Univ. of (1999) used an analytical solution to determine greenfield tunneling
Nottingham, University Park, Nottingham NG7 2RD, United Kingdom. displacements and a boundary element analysis to evaluate their
E-mail: alec.marshall@nottingham.ac.uk
effect on adjacent piles. As discussed earlier, this assumption may
Note. This manuscript was submitted on August 2, 2011; approved on
February 7, 2012; published online on February 9, 2012. Discussion period be valid for bored piles; however, its applicability to jacked or driven
open until March 1, 2013; separate discussions must be submitted for piles is not certain because of the use of greenfield displacements as
individual papers. This paper is part of the Journal of Geotechnical and the input for the solution. Poulos and Deng (2004) used standard
Geoenvironmental Engineering, Vol. 138, No. 10, October 1, 2012. bearing-capacity equations and soil-pile skin friction coefficients to
©ASCE, ISSN 1090-0241/2012/10-1237–1246/$25.00. evaluate the load-carrying capacity of piles and then estimated the

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / OCTOBER 2012 / 1237


reduction in pile capacity using an elastic, constant volume cavity
contraction analysis.
This paper considers the problem of tunnel construction beneath
end-bearing piles and provides an analytical method for evaluating
the safe relative tunnel-pile distance based on an expected tunnel
volume loss. The results of the method are compared against data
obtained from centrifuge model experiments of tunnel volume loss
beneath piles jacked in sand. The analysis is based on the limit
pressure spherical cavity expansion method set out in Randolph et al.
(1994) to evaluate the end-bearing capacity of a pile and ground
stresses caused by pile installation. The Randolph et al. (1994) frame-
work calculations are also used to estimate an average mobilized
skin-friction shear stress during pile installation. The next stage of
the analysis is similar to that provided by Poulos and Deng (2004), Fig. 1. Influence zones defined by Jacobsz et al. (2004)
who used an elastic, constant-volume cylindrical cavity contraction
analysis to estimate the reduction in confining stress at the location of
the pile so that a reduced pile capacity could be calculated. In this
analysis, however, the soil is treated as elastic-perfectly plastic Centrifuge Experiments
(using the Mohr-Coulomb failure criterion) and includes soil di-
lation. In addition, a modified soil modulus is used to account for the In a subsequent section, the analysis studies the centrifuge experi-
effect of pile installation on soil stiffness in the region between ment data provided by Jacobsz (2002) and Marshall (2009). This
the tunnel and pile (based on the ground stresses calculated from the section provides a description of the centrifuge models. Additional
initial spherical cavity expansion analysis used to evaluate pile end- information on the centrifuge model equipment and the tests per-
bearing resistance). The method presented in this paper does not formed can also be found in Jacobsz et al. (2004), Marshall et al.
attempt to estimate the effect of tunneling on skin friction. Instead, (2010), and Marshall and Mair (2011).
the calculated mobilized shaft skin friction from pile installation is Centrifuge modeling involves testing of scaled-down models of
assumed to remain constant during tunnel volume loss. This sim- full-sized prototypes within an elevated gravity environment (Ng,
plifying assumption is supported by experimental data from Jacobsz where g is gravity and N is a scaling factor). Centrifuge modeling
(2002), who concluded that tunnel volume loss did not have a sig- provides an effective method of studying soil-structure interaction
nificant effect on shaft capacity of piles directly above a tunnel and problems because it realistically replicates the nonlinear soil and soil-
that disregarding the increase in normal stress around the pile shaft is structure interaction behavior. Centrifuge modeling has provided
conservative. valuable data for the analysis of tunneling induced soil displacements
(Mair et al. 1993; Marshall et al. 2012), as well as for soil-structure
interaction problems related to tunneling (Vorster et al. 2005; Marshall
Overview of Analysis et al. 2010; Marshall and Mair 2011).
A schematic of the Marshall (2009) centrifuge model is presented
The general philosophy of the analysis in this study is to provide in Fig. 2(a). The tests were performed at a centrifuge acceleration of
a relatively fast method to determine the approximate safe loca- N 5 75, representing a prototype scale tunnel diameter, Dt, of 4.65 m
tion of tunnel construction beneath end-bearing piles without the buried with its axis at a depth, zt, of 13.65 m. The piles, which were
need for computationally expensive and time-consuming finite- half cylindrical in shape so that they could be placed against the
element or finite-difference analyses. The method is based on the Perspex wall of the centrifuge strong-box [Fig. 2(b)], had a proto-
combination of (1) spherical cavity expansion analysis to de- type diameter, Dp of 0.9 m. The experimental setup allowed for the
termine the end-bearing capacity of the pile and (2) cylindrical acquisition of digital images of the soil and piles through the Perspex
cavity contraction to estimate the change in confining stress within wall during the tests. The images were then analyzed using geoPIV
the soil and the resulting reduction in load-carrying capacity of (White et al. 2003) to determine subsurface soil and pile dis-
the pile caused by tunnel volume loss, including a modification placements. The piles were pushed into the ground by approximately
of soil stiffness to account for the effect of pile installation. two pile diameters at 75g, after which the load on top of the piles was
The method relies on the assumption that changes in stress reduced to roughly 60% of the maximum jacking force. Water was
from 2 can be superimposed on the stress state caused by 1. This then extracted from the model tunnel to simulate volume loss, and
simplifying assumption, based on elastic soil behavior, is not the displacement of the piles was measured using geoPIV and linear
strictly correct because the analyses do account for soil yielding, variable differential transformers (LVDTs) on the pile caps. The
adopting a Mohr-Coulomb yield criterion. Nevertheless, the re- service load applied to the piles was kept constant during tunnel
sults of the analysis provide rational results when compared with volume loss. Two centrifuge experiments were conducted by
data obtained from centrifuge tests of jacked end-bearing piles Marshall (2009), labeled TP1 and TP2, with different offsets from
above tunnels in sand provided by Jacobsz (2002) and Marshall the tunnel to the piles [Figs. 2(c and d)].
(2009). The Jacobsz (2002) experiments were conducted using the same
The analysis considers driven or jacked end-bearing piles with sand and model pile size as Marshall (2009), as well as similar
their tips located within the influence zones defined by Jacobsz et al. equipment to replicate tunnel volume loss. In the Jacobsz experi-
(2004) (Fig. 1), where there is a potential for large pile displacements ments, however, the piles were placed within the middle of the
to occur as a result of tunnel construction. As discussed previously, centrifuge container and were fully cylindrical. In addition, the
the effect of tunnel volume loss on pile shaft friction is not con- ultimate pile loads were reduced to roughly 50% of maximum
sidered in this analysis. A constant value of average mobilized soil- jacking loads. It should be noted, however, that Jacobsz (2002)
pile skin shear stress is assumed, which is based on the framework varied this value between 50 and 82% and noted no discernible
equations provided by Randolph et al. (1994). effect on the response of the piles to tunnel volume loss. In prototype

1238 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / OCTOBER 2012


Fig. 2. Centrifuge model test details for Marshall (2009)

scale, the Jacobsz tunnel was 4.5 m in diameter and located at a depth was calculated as p90 5 2s9y =3, based on an assumption of an at-rest
to axis level of 21.5 m. lateral earth pressure coefficient, K0, of 0.5 and where s9y is vertical
earth pressure.
Also from Randolph et al. (1994), the small strain shear modulus,
Pile End-Bearing Capacity Using Spherical G0, was used as the representative value of shear stiffness and was
Cavity Expansion estimated using Eq. (3) as follows:
 n
This part of the analysis follows the logic set out by Randolph et al. G0 p9
(1994), in which the spherical cavity expansion method is used to ¼ S exp ðc1 Id Þ 0 ð3Þ
pa pa
estimate the end-bearing capacity of piles. The analysis requires an
input of the mean effective stress, p90 ; the friction angle, f9; the where S 5 600; c1 5 0.7; n 5 0.43 (Lo Presti 1987); and pa 5
dilation angle, c; shear stiffness, G; and Poisson’s ratio, n. atmospheric pressure (100 kPa).
For this analysis, the in situ mean effective stress p90 was taken at The spherical cavity expansion analysis is based on an isotropic
the depth of the pile tip. As discussed in Randolph et al. (1994), for dilatant elastic-perfectly plastic material with a nonassociated
a closed-ended pile, the values of friction and dilation angles adopted Mohr-Coulomb yield criterion. The analysis allows estimation of
in the analysis should be the average of the initial (maximum) values the limiting end-bearing pressure to cause failure beneath the pile,
and the final (critical state) values; hence, f9 5 0:5ðf9max 1 f9cy Þ and p9lim , as well as the associated distribution of stress within the soil.
c 5 0:5cmax , where f9cy is the critical state friction angle. The work This part of the analysis is based on the expansion of cavities in an
of Bolton (1986, 1987) can be used to obtain an estimate of infinite medium, derived within Yu and Houlsby (1991) and Yu
the peak friction and dilation angles in triaxial strain conditions: (2000). The necessary parameters [Eq. (20)] are from Yu (2000) and
f9max 2 f9cy 5 cmax 5 3IR , where IR is the relative dilatancy index. have been reproduced here so that the reader may follow the cal-
The value of IR is dependent on relative density, Id, and confining culation sequence more readily. The reader may refer to Yu (2000)
pressure, p9, as follows: IR 5 5Id 2 1 for p9 # 150 kPa, and for the complete derivation of the equations. The parameter k is used
IR 5 Id ½5 2 lnðp9=150Þ 2 1 for p9 . 150 kPa (Bolton 1987). Using to differentiate between spherical (k 5 2) and cylindrical (k 5 1)
the values of f9max and cmax from Bolton (1986, 1987) to determine analyses.
f9 and c gives the following Eqs. (1) and (2), respectively: Analytically, the pressure required to expand a cavity from
zero radius to a finite radius is equivalent to that required to ex-
f9 ¼ f9cy þ 1:5IR ð1Þ pand a cavity from a finite radius to infinity. The solution pre-
sented here to determine p9lim is the latter, although both methods
c ¼ 1:5IR ð2Þ are provided in Yu (2000). The limit pressure p9lim is found using
the Eqs. (4)–(6) by varying the value of p9lim in Eq. (4) until the
The value of Id was 90% for the Marshall (2009) tests and varied left and right sides of Eq. (5) are equal; the summation on the left
between 76 and 79% for the Jacobsz (2002) tests. The value of f9cy side of Eq. (5) was found to converge quickly (n 5 10 was
for the soil was taken as 32° for both cases. The mean effective stress sufficient):

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / OCTOBER 2012 / 1239


 
ðk þ aÞ Y þ ða 2 1Þp9lim
Rlim ¼ ð4Þ
að1 þ kÞ½Y þ ða 2 1Þp90 

P
‘ x b1k
An ðRlim ; mÞ ¼ ð1 2 dÞ b ð5Þ
n¼0 g

8 n
>
> m
< ln R if n ¼ g
n!
An ðR; mÞ ¼ mn ð6Þ
>
> ðRn2g 2 1Þ
: otherwise
n!ðn 2 gÞ

The relationship between the end-bearing pressure of the pile, qb,


and p9lim is given by Eq. (7) (Randolph et al. 1994) as follows:
h   i
qb ¼ p9lim 1 þ tan fcvA
9 tanðac Þ ð7Þ

Fig. 3. Ground stresses at end of pile jacking based on cavity expansion


where ac 5 45 1 f9cy =2. analysis of centrifuge test TP2
For the experimental data of Marshall (2009) and Jacobsz (2002),
the model piles had conical tips (45 and 60° angles, respectively).
These angles were therefore used in Eq. (7) in place of ac, and an
To account for the effect of pile installation on the increased
interface friction angle, ds, of ðf9cy 2 5°Þ was used in place of
stiffness of the soil surrounding the pile, an increased value of shear
f9cvA [resulting in ds 5 27°, which is in agreement with estimates of
stiffness, Gmod, was used. This was done rather arbitrarily by adding
interface friction made by Jacobsz (2002)].
the value of p9 determined from the increase in ground stresses
resulting from pile installation at the midpoint between the pile
tip and the tunnel boundary to the isotropic value of p90 calculated
Ground Stresses Attributable to Pile Jacking
at the depth of the tunnel. The value of Gmod was then calculated
using this increased value of p9 in Eq. (3). All other parameters
A spherical zone of yielded soil is created around the tip of the pile
were recalculated based on the isotropic p90 value. The result of
during jacking. The ground stress profile within the yielded plastic
increasing stiffness in this step of the analysis is that confining
zone and the nonyielded elastic zone can be evaluated using cavity
stresses at the pile tip are reduced more significantly with volume
expansion methods. Using p9lim (from the previous section) as the
loss, thus providing a more conservative approach. A quantitative
cavity pressure, p, the distance, c, from the pile tip to the elastic-
evaluation of the effect of this process is given subsequently.
plastic interface can be evaluated using Eq. (8) (Yu 2000) as follows:
For an assumed percentage value of tunnel volume loss, Vl,
 a which is based on the assumption of concentric contraction, the
c ðk þ aÞ½Y þ ða 2 1Þp kða21Þ tunnel radius after volume loss can be calculated using Eq. (9) as
¼ ð8Þ
a0 að1 þ kÞ½Y þ ða 2 1Þp90  follows:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where a0 5 original radius of the cavity and is taken as the radius of a ¼ a0 ð1 2 Vl =100Þ ð9Þ
the pile. The ground stresses in the elastic and plastic zones are then
calculated using Eq. (21), and the confining stress is calculated as For very low volume losses, the soil around the tunnel may not
p9 5 ðsr 1 2su Þ=3. The effect of pile unloading is not considered in yield. and tunnel pressure, radial stress, and circumferential stress
this analysis for the reason that ground stresses are generally locked within the soil can be calculated using Eq. (22). As volume loss
into the ground after pile jacking. This is supported by observations increases and tunnel pressure decreases, the soil around the tunnel
by Jacobsz (2002) that pile response to tunnel volume loss did not will eventually yield. Analytically, the soil around the tunnel will
vary for piles unloaded to 50–82% of their maximum jacking loads. yield when tunnel pressure becomes less than p1y, given by Eq. (10)
An example of ground stresses determined from the cavity ex- (Yu 2000) as follows:
pansion analysis for centrifuge test TP2 is shown in Fig. 3.
ð1 þ kÞ ðkYÞ
p1y ¼ p0 2 ð10Þ
ð1 þ akÞ ð1 þ akÞ
Effect of Tunnel Volume Loss on Ground Pressure
After yield, the procedure for determining ground stresses is
The analysis now considers the effect of tunnel volume loss on the now described. First, the ratio of c0/c, where c0 is the original radius
confining pressure at the location of the pile tip. The case of con- to the elastic-plastic interface and c is the new radius, is determined
centric contraction of a cylinder from an in situ stress state is con- by Eq. (11) (Yu 2000) as follows:
sidered. The cavity contraction analysis and equations presented
here are derived from Yu (2000). It is assumed that an isotropic c0 ð1 2 aÞp0 þ Y
¼ 1þ ð11Þ
pressure, p90 , acts throughout a homogeneous soil. The value of p90 c 2ð1 þ kaÞG
is calculated at the depth of the tunnel axis. Initially, the tunnel has
a radius of a0 and an internal pressure, p0, equal to the isotropic Using Eq. (12) (Yu 2000), the value of tunnel pressure, p, can be
pressure of the soil. The analysis is for a cylindrical cavity and iterated until the calculated value of a matches that specified by the
therefore k 5 1. value of volume loss [calculated by Eq. (9)] as follows:

1240 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / OCTOBER 2012


0  11kb  1
c
ln 1 2 1 2 X 0
a0 B c C
¼ expB
@
C
A
a 1 þ kb
ð12Þ
 11kb
ð1 þ kaÞ½Y þ ða 2 1Þp kð12aÞ
X ¼
ð1 þ kÞ½Y þ ða 2 1Þp90

Once the tunnel pressure, p, is determined, the ratio between


the plastic radius and the tunnel radius, c/a, can be determined using
Eq. (13) (Yu 2000) as follows:
 1
c ð1 þ kaÞ½Y þ ða 2 1Þp kð12aÞ
¼ ð13Þ
a ð1 þ kÞ½Y þ ða 2 1Þp90

With the knowledge of c, the postyield ground stresses within


the elastic and plastic zones around the contracting tunnel can be
determined using Eq. (23).
Fig. 4. Contour of Rp for centrifuge test TP2
Effect of Tunnel Volume Loss on Pile
Load-Carrying Capacity displacements [0.5–2% of pile diameter according to Fleming et al.
(1992)], it can reasonably be assumed that the full shaft capacity is
The effect of tunnel volume loss on the end-bearing capacity of the mobilized on small pile movements caused by tunnel volume loss. In
piles can now be estimated by evaluating the change in confining addition, Jacobsz (2002) concluded that stress changes within the
stress at the location of the pile tip because of tunnel volume loss, soil did not have a significant effect on shaft capacity for piles
dp9. Given the assumption of isotropic pressure acting throughout directly above the tunnel and that disregarding the increase in normal
the soil for the cavity contraction analysis from the previous section, stresses along the pile shaft that results from tunnel volume loss is
the initial confining pressure is equal to p90 . The values of su and sr conservative. The Randolph et al. (1994) method should therefore
are calculated at the location of the pile tip using Eqs. (22) or (23) provide a rational estimate of pile shaft capacity and the simplifying
depending on the magnitude of volume loss and determined state of assumption that volume loss does not affect shaft capacity can be
soil. The cylindrical cavity contraction model assumes plane strain applied as a conservative approach.
conditions during unloading and therefore the change in out-of- Following Randolph et al. (1994), the maximum shear stress
plane stress, dsy, can be calculated as dsy 5 n(dsu 1 dsr) (Yu and along the pile shaft, tmax can be calculated using Eq. (16) as follows:
Houlsby 1991). The change in confining pressure at the location of
the pile tip can therefore be calculated as dp9 5 [(1 1 n)(dsr 1 t max ¼ qb St tanðds Þ ð16Þ
dsu)]/3.
A confining stress reduction factor, Rp (Fig. 4 illustrates a contour where St 5 a exp [2b tan(fcy)]; a 5 2; b 5 7; and the values of qb
of Rp for data from centrifuge test TP2) can be calculated using Eq. and ds are from the previous discussion. The distribution of bs along
(14) as follows: the pile length (a subscript s is used here to differentiate between the
b term used in the cavity expansion equations in the Appendix),
dp9
Rp ¼ 1 2 ð14Þ where bs 5 t s =s9y was calculated using Eq. (17) (Randolph et al.
p90 1994) as follows:
The effect of tunnel volume loss on the end-bearing capacity of  
bs ðzÞ ¼ bmin þ ðbmax 2 bmin Þexp 2 ms ðL 2 zÞ=Dp ð17Þ
the pile is then determined by applying the reduction factor from
Eq. (14) to the value of confining stress, p90 , used in the initial bearing where bmin was taken as 0.2, bmax 5 StNq tan(ds); Nq 5 qb =s9y (s9y at
capacity analysis and then repeating this analysis to determine a new pile tip); ms 5 a parameter that controls the rate of decay of bs along
value of end-bearing capacity after tunnel volume loss: qb,yl. the pile shaft and was taken as 0.05; and z 5 depth measured from the
A reduction factor for the end-bearing capacity can be defined as surface. The distribution of shear stress along the pile shaft is then
Rq 5 qb,yl/qb. This term, however, does not include the effect of pile t s ðzÞ 5 bs ðzÞs9y ðzÞ.
length, which limits the ability to compare results from different piles. An equivalent average shear stress, ts , which provides the same
The use of total pile load capacity, Q, does allow comparison of shaft load as the distributed shaft shear stress, can be calculated by
results from piles of different size. The value of Q is determined as integrating the shear stresses along the pile length [Eq. (18)]. This
the summation of the contribution from the pile base, Qb, and the value of t s can then be used in Eq. (15) to determine the total load-
shaft, Qs as follows: carrying capacity of the pile as follows:
.
Q ¼ Qb þ Qs ¼ qb pD2p 4 þ t s pDp L ð15Þ ÐL
t s dz
ts ¼ z ¼ 0 ð18Þ
L
where t s 5 average shear stress along the pile shaft (skin friction);
Dp 5 pile diameter; and L 5 embedded length of the pile (Fleming Using this method, the calculated values of t s were roughly 57 kPa
et al. 1992). for the Marshall (2009) tests and 93 kPa for the Jacobsz (2002) tests.
The framework equations of Randolph et al. (1994) can also A pile capacity reduction factor, RQ, is defined in the following
be used to evaluate an average skin friction, t s , based on pile equation to compare the fraction of pile capacity at failure because
jacking. As shaft shear stress is mobilized after relatively small pile of tunnel volume loss, Qyl, against the initial capacity, Qo:

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / OCTOBER 2012 / 1241


.
Qyl qb;yl pD2p 4 þ t s pDp L qb;yl Dp þ 4t s L
RQ ¼ ¼ . ¼ (19)
Qo qb pD2p 4 þ t s pDp L qb Dp þ 4t s L

This normalization considers the size and embedment of the piles


and therefore allows comparison of data between piles of different
size and depth.

Comparison with Centrifuge Test Results

The results of the analysis of the Jacobsz (2002) and Marshall


(2009) centrifuge tests are shown in Fig. 5. Fig. 5(a) (from Marshall
and Mair 2011) plots experimental data of the relationship between
the normalized relative tunnel-pile tip position ðdt2p 2
=rt2 Þ and the
volume loss at which failure occurred. For the interpretation of the
centrifuge test data, pile failure was taken as the moment when
the rate of change of the pile displacement showed a significant
increase. Fig. 5(b) plots the values of RQ [Eq. (19)] calculated using
the analysis method presented in the previous sections for all of the
centrifuge experiments presented in Fig. 5(a). The solid data points
indicate piles within Zone A of Fig. 1 (directly above the tunnel),
whereas the open data points indicate piles within Zone B.
As discussed in Marshall and Mair (2011), the data in Fig. 5(a)
suggest that there is a relationship between the volume loss at
which pile failure occurs and the square of the relative distance
between the tunnel and pile tip. Fig. 5(b) shows that the value of RQ
is between 0.69 and 0.77 at pile failure. This means that at the
moment of pile failure, the pile capacity was reduced to between
69 and 77% of its original value as a result of the reduction in confining
stress at the pile tip because of tunnel volume loss (based on this
analysis and the assumptions associated with RQ discussed previously).
It should be noted that, except for one data point, the experimental
results are based on piles with an applied service load during tunnel Fig. 5. Results of analysis for centrifuge tests
volume loss that was between 50 and 60% of the maximum jacking
load. However, Jacobsz (2002) showed that similar pile behavior
was measured for piles with substantially different service loads Discussion
(ranging between 53 and 82% of the maximum jacking load). Indeed,
one of the data points included in Fig. 5 is from the pile with a service Fig. 6 illustrates how the analysis can be used to estimate the critical
load of 82% of the maximum jacking load (RQ 5 0.72, Vl 5 1.4%). volume loss at which pile capacity is reduced to a given value of RQ.
Two data points available from Jacobsz (2002) and one from As the value of RQ is decreased, the analysis results logically show
Marshall (2009) have been omitted from Fig. 5 even though the tips that the allowable volume loss increases. These results are based on
of the piles were located within the critical influence zones identified the reference parameters provided in Table 1 and assuming a closed-
in Fig. 1, where large pile displacements are expected. These piles ended, flat-based pile.
failed at unexpectedly high volume losses. For the omitted Jacobsz An evaluation of the sensitivity of the analysis to the variation of
piles, it is likely that the failure was because the pile tips were key material parameters was performed, the results of which are
located very close to the interface to Zones C and D in Fig. 1. Despite shown in Fig. 7. The variation of parameters is indicated within each
the fact that the tips were located within the critical influence zone, of the subplots in Fig. 7. Results are presented in a similar form as
the soil beneath the tips would have been significantly compacted Fig. 5(a), where normalized distance from the tunnel axis to the pile
during the pile jacking stage, resulting in a densified bulb of soil with tip is plotted against volume loss. Note, however, that the curves
high strength and stiffness (White and Bolton 2004; Marshall 2009; illustrate the volume loss required to obtain RQ 5 0.85. This was
Marshall and Mair 2011). This bulb of densified soil could have done by iterating volume loss and dt2p within the analysis until the
transferred the pile load to the soil within Zone D where tunnel desired value of RQ was obtained. The data presented in Fig. 6 are
volume loss does not have a significant effect on pile-bearing ca- based on the geometric parameters given in Table 1 and would vary
pacity. For the omitted Marshall pile, it was observed that, upon ini- for other geometric conditions.
tiation of the failure of Pile 1 in test TP2 (refer to Fig. 2), displacements The strength parameters f9cy and C in Figs. 6(a and b) have minor
were concentrated in the zone between the tunnel and Pile 1 and that effects on the results because they contribute to both qb,yl and qb.
very little disturbance was observed in the region between the tunnel However, the magnitude of contribution to each of these is different
and Pile 2. The effect of tunnel volume loss was then concentrated on for a given increase in strength parameter. For an increase of f9cy ,
Pile 1, resulting in failure of Pile 2 at a very high volume loss. the results show a small increase in the critical value of volume loss
The data presented in Fig. 5(b) suggests that a conservative value (Vl,crit) necessary to achieve RQ 5 0.85 (a beneficial effect). The
of RQ for the experimental data are 0.85. The effect of the variation results for C, however, show an opposite effect, with a small de-
of key soil parameters on the results of the analysis of the centrifuge crease in Vl,crit noted for an increase in C. This is because, for a given
experiments is presented in the next section. increase in C, the value of qb increases slightly more than that of

1242 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / OCTOBER 2012


Gmod/Gtun decreases exponentially as the magnitude of tunnel-pile
separation increases. This is because Gmod is influenced by the
stresses determined from the pile jacking process that are based on
the spherical cavity expansion analysis.

Conclusions

This study presented a method to evaluate the effect of tunnel con-


struction on the load-carrying capacity of end-bearing piles. The
method is based on calculation of pile end-bearing capacity using
a spherical cavity expansion analysis and estimation of the reduction of
confining pressure at the pile tip that results from tunnel volume loss
using a cylindrical cavity contraction analysis. The reduced confining
pressure is then used to reevaluate the pile end-bearing capacity, and
Fig. 6. Analysis results solving for volume loss necessary to obtain the reduction of total pile capacity is determined using an average
values of RQ shear stress along the pile shaft. A modified shear modulus is included
to account for the effect of pile installation on shear stiffness.
Table 1. Reference Parameters for Parametric Analysis The analytical results were compared against centrifuge test data,
and it was shown that a reduction factor, RQ, of 0.85 conservatively
Parameter Symbol Unit Value predicted the volume losses at which pile failure occurred for the
Tunnel depth zt m 15 analyzed tests. It should be noted, however, that the applicability
Tunnel radius rt m 5 of this value of RQ has not been validated against field data, and
Pile radius rp m 1 therefore, its general use should be applied with appropriate judg-
Normalized pile tip depth zp /(zt 2 rt) — 0.8 ment. Results of a parametric analysis were presented that illustrate
Critical state friction angle f9cy ° 30 the variation of analysis results for a range of material properties.
Cohesion C kPa 0
Unit weight g kN/m3 18 Appendix. Cavity Expansion/Contraction Equations
Relative density Id — 0.8
At rest earth pressure K0 — 0.5 The equations provided within this appendix were obtained from Yu
coefficient (2000).
Poisson’s ratio n — 0.2 The following parameters are used for cavity expansion and cavity
contraction analyses:
qb,yl in the analysis. This would result in a decrease in RQ, so to E E 2C cosf
obtain RQ 5 0.85, a higher value of normalized tunnel-pile G¼  M¼  Y¼  
2ð1 þ nÞ 12n2 ð22kÞ 12sinf
separation ðdt2p =rt Þ2 is necessary, resulting in a shift upward
of the curves in Fig. 6(b) as C increases. Figs. 6(c and d) show that 1 þ sinf 1 þ sinc aðb þ kÞ
an increase in relative density, Id, has a detrimental effect (de- a¼ b¼  g ¼  
12sinf 12sinc kða21Þb
creasing the value of Vl,crit), whereas unit weight has a beneficial  
effect. The effect of K0 and Poisson’s ratio in Figs. 6(e and f) are Y þ ða21Þp90 ð1 þ kÞd 12n2 ð22kÞ
d¼  m ¼
shown to be quite small. 2ðk þ aÞG ð1 þ nÞða21Þb
As described previously, the effect of pile installation on soil 
stiffness was taken into account when evaluating the effect of tunnel knða þ bÞ
 ab þ kð122nÞ þ 2n2  
volume loss on the confining stress at the pile tip location. This was 12nð22kÞ
done by using an increased value of shear stiffness, Gmod, based on 
ðb þ kÞð122nÞð1 þ ð22kÞnÞ½Y þ ða21Þp0 
an increase in confining stress at the midpoint between the tunnel x ¼ exp
boundary and the pile tip. Fig. 8(a) illustrates the effect of using Gmod Eða21Þb
compared with results obtained when stiffness was not modified ð20Þ
(labeled Gtun and calculated based on p90 at the depth of the tunnel
axis). The Gmod method is more conservative because it provides The following are the ground stresses resulting from pile jacking
a lower value of Vl,crit than using Gtun. Fig. 8(b) shows the ratio of (superscripts e and p refer to elastic and plastic, respectively):

Plastic zone: r , c Elastic zone: r . c


Y
ser ¼ 2 p90 2 Br 2ð11kÞ
kða21Þ
spr ¼ þ Ar2 a
a21
Y A kða21Þ B
spu ¼ þ r2 a seu ¼ 2 p90 þ r 2ð11kÞ
a21 a k
   
ð1 þ kÞa Y þ ða 2 1Þp90 kða21Þ k Y þ ða 2 1Þp90 11k
A ¼ 2 c a B ¼ c ð21Þ
ða 2 1Þðk þ aÞ k þa

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / OCTOBER 2012 / 1243


Fig. 7. Results of material property parametric analysis

11k
Preyield ground stresses resulting from tunnel volume loss are as sr ¼ 2 p90 2 ð p 2 p90 Þ a  
r
follows:  
p 2 p90 a 11k
su ¼ 2 p90 2 ð22Þ
k r
2kGða0 2 aÞ Postyield ground stresses resulting from tunnel volume loss are as
p ¼ p90 þ  
a follows:

Elastic zone: r . c Plastic zone: r , c


ser ¼ 2 p0 2 Br 2ð11kÞ spr ¼ Y þ Arkða21Þ
a21
seu ¼ 2 p0 þ Br2ð11kÞ spu ¼ Y þ Aar kða21Þ cð12aÞk ð23Þ
 k a21  
k ð1 2 aÞp90 2 Y 11k ð1 þ kÞ Y þ ða 2 1Þp90
B ¼ c A ¼ 2
1 þ ka ða 2 1Þð1 þ kaÞ

Acknowledgments Notation

The author would like to acknowledge financial support from the The following symbols are used in this paper:
Natural Sciences and Engineering Research Council of Canada An 5 cavity expansion analysis parameter used to determine
(NSERC) and the Cambridge Commonwealth Trust—Kenneth p9lim ;
Sutherland Memorial Scholarship. Ap 5 cross-sectional area of pile;

1244 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / OCTOBER 2012


Qb 5 pile shaft contribution to total pile load capacity;
Qyl 5 total pile load capacity at failure because of tunnel
volume loss;
Q0 5 initial total pile load capacity;
qb 5 pile end-bearing pressure;
qb,yl 5 new value of pile end-bearing pressure after tunnel
volume loss;
Rlim 5 elastic-plastic interface radius at limiting pressure;
Rp 5 reduction factor—confining stress;
RQ 5 reduction factor—pile load capacity;
Rq 5 reduction factor—bearing capacity;
rt 5 tunnel radius;
S 5 parameter used to determine G0;
St 5 ratio of radial effective stress near pile tip at failure to qb;
Vl 5 tunnel volume loss, in %;
Y 5 cavity expansion analysis parameter related to C and f;
z 5 depth, measure from surface;
zt 5 depth to tunnel axis, measure from surface;
a 5 cavity expansion analysis parameter related to f;
ac 5 parameter used to determine qb;
b 5 cavity expansion analysis parameter related to c;
bs 5 ratio of pile shaft shear stress to vertical stress within
soil ðts =s9y Þ;
bmax 5 maximum value of bs;
Fig. 8. Effect of Gmod bmin 5 minimum value of bs;
x 5 cavity expansion analysis parameter;
d 5 cavity expansion analysis parameter;
a5 current cavity radius; ds 5 interface friction angle;
a0 5 original cavity radius; f9 5 friction angle;
C5 cohesion; f9 5 average friction angle;
c5 current distance from pile tip to elastic-plastic interface; f9cy 5 critical state (constant volume) friction angle;
c0 5 original distance from pile tip to elastic-plastic f9max 5 maximum friction angle;
interface; g 5 cavity expansion analysis parameter;
c1 5 parameter used to determine G0; m 5 cavity expansion analysis parameter;
Dp 5 pile diameter; ms 5 parameter that controls the rate of decay of bs along the
Dt 5 tunnel diameter; pile shaft;
dt2p 5 relative distance between tunnel axis and pile tip; y 5 Poisson’s ratio;
E5 Young’s modulus; s9r 5 radial stress;
G5 shear stiffness/modulus; s9y 5 vertical earth pressure;
Gmod 5 modified soil stiffness to account for effect of pile s9u 5 circumferential stress;
installation; tmax 5 maximum shear stress along pile shaft;
Gtum 5 shear stiffness calculated based on tunnel axis depth; ts 5 shear stress along pile shaft;
G0 5 small strain shear stiffness/modulus; t s 5 average shear stress along pile shaft;
g5 gravity; c 5 dilation angle;
Id 5 relative density; c 5 average dilation angle; and
IR 5 relative dilatancy index; cmax 5 maximum dilation angle.
K0 5 at-rest lateral earth pressure coefficient;
k5 cavity expansion parameter 5 spherical k 5 2;
cylindrical k 5 1;
References
L5 embedded length of pile;
N5 centrifuge scaling factor;
Attewell, P. B., Yeates, J., and Selby, A. R. (1986). Soil movements induced
n5 parameter used to determine G0; by tunnelling and their effects on pipelines and structures, Blackie and
p5 cavity pressure; Son, London.
pa 5 atmospheric pressure (100 kPa); Bezuijen, A., and Van der Schrier, J. (1994). “The influence of a bored tunnel
p9lim 5 limiting end-bearing pressure; on pile foundations.” Proc., Int. Conf., Centrifuge ’94, C. F. Leung,
p0 5 initial tunnel pressure; F. Lee, and T. Tan, eds., Balkema, Singapore, 681–686.
p90 5 mean effective stress; Bolton, M. D. (1986). “The strength and dilatancy of sands.” Geotechnique,
36(1), 65–78.
p1y 5 pressure at which soil yields around a contracting Bolton, M. D. (1987). “Discussion on the strength and dilatancy of sands.”
cylinder; Geotechnique, 37(2), 219–226.
Q5 total pile load capacity; Chen, L. T., Poulos, H. G., and Loganathan, N. (1999). “Pile responses
Qb 5 pile base contribution to total pile load capacity; caused by tunneling.” J. Geotech. Geoenviron. Eng., 125(3), 207–215.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / OCTOBER 2012 / 1245


Cheng, C. Y., Dasari, G. R., Chow, Y. K., and Leung, C. F. (2007). Marshall, A. M., Klar, A., and Mair, R. J. (2010). “Tunneling beneath buried
“Finite element analysis of tunnel-soil-pile interaction using dis- pipes—A view of soil strain and its effect on pipeline behavior.”
placement controlled model.” Tunn. Undergr. Space Technol., 22(4), J. Geotech. Geoenviron. Eng., 136(12), 1664–1672.
460–466. Marshall, A. M., and Mair, R. J. (2011). “Tunneling beneath driven or jacked
Fleming, W., Weltman, A., Randolph, M. F., and Elson, W. (1992). Piling end-bearing piles in sand.” Can. Geotech. J., 48(12), 1757–1771.
engineering, 2nd Ed., Blackie and Son, Glasgow, U.K. Meguid, M. A., and Mattar, J. (2009). “Investigation of tunnel-soil-pile
Franzius, J. N., Potts, D. M., and Burland, J. B. (2006). “The response of interaction in cohesive soils.” J. Geotech. Geoenviron. Eng., 135(7),
surface structures to tunnel construction.” Proc. Inst. Civil Eng. Geotech. 973–979.
Eng., 159(1), 3–17. Morton, J. D. and King, K. H. (1979). “Effects of tunnelling on the bearing
Jacobsz, S. W. (2002). “The effects of tunnelling on piled foundations.” capacity and settlement of piled foundations.” Proc. Tunnelling ’79,
Ph.D. thesis, Cambridge Univ., Cambridge, U.K. Institute of Mining and Metallurgy, London, 57–68.
Jacobsz, S. W., Standing, J. R., Mair, R. J., Hagiwara, T., and Sugiyama, T. Mroueh, H., and Shahrour, I. (2002). “Three-dimensional finite element
(2004). “Centrifuge modelling of tunnelling near driven piles.” Soil analysis of the interaction between tunneling and pile foundations.” Int.
Found., 44(1), 49–56. J. Numer. Anal. Methods Geomech., 26(3), 217–230.
Kitiyodom, P., Matsumoto, T., and Kawaguchi, K. (2005). “A simplified Potts, D. M., and Addenbrooke, T. I. (1997). “A structure’s influence on
analysis method for piled raft foundations subjected to ground movements tunneling-induced ground movements.” Proc. Inst. Civil Eng., Geotech.
induced by tunnelling.” Int. J. Numer. Anal. Methods Geomech., 29(15), Eng., 125(2), 109–125.
1485–1507. Poulos, H. G., and Deng, W. (2004). “An investigation on tunnelling-
Klar, A., and Marshall, A. M. (2008). “Shell versus beam representation of induced reduction of pile geotechnical capacity.” Proc., 9th Australia
pipes in the evaluation of tunneling effects on pipelines.” Tunn. Undergr. New Zealand Conf. on Geomechanics, Vol. 1, New Zealand, Geo-
Space Technol., 23(4), 431–437. technical Society & Australian Geomechanics Society, Auckland,
Klar, A., Vorster, T. E. B., Soga, K., and Mair, R. J. (2005). “Soil-pipe New Zealand, 116–122.
interaction due to tunnelling: Comparison between Winkler and elastic Randolph, M. F., Dolwin, J., and Beck, R. (1994). “Design of driven piles
continuum solutions.” Geotechnique, 55(6), 461–466. in sand.” Geotechnique, 44(3), 427–448.
Lee, C. J., and Jacobsz, S. W. (2006). “The influence of tunnelling on Vorster, T. E. B., Klar, A., Soga, K., and Mair, R. J. (2005). “Estimating the
adjacent piled foundations.” Tunn. Undergr. Space Technol., 21(3–4), effects of tunneling on existing pipelines.” J. Geotech. Geoenviron. Eng.,
430. 131(11), 1399–1410.
Lo Presti, D. (1987). “Mechanical behaviour of ticino sand from resonant White, D. J., and Bolton, M. D. (2004). “Displacement and strain paths
column tests.” Ph.D. thesis, Politecnico di Torino, Torino, Italy. during plane-strain model pile installation in sand.” Geotechnique, 54(6),
Mair, R. J., Taylor, R. N., and Bracegirdle, A. (1993). “Subsurface 375–397.
settlement profiles above tunnels in clays.” Geotechnique, 43(2), White, D. J., Take, W. A., and Bolton, M. D. (2003). “Soil deformation
315–320. measurement using particle image velocimetry (PIV) and photogram-
Marshall, A. M. (2009). “Tunnelling in sand and its effect on pipelines and metry.” Geotechnique, 53(7), 619–631.
piles.” Ph.D. thesis, Cambridge Univ., Cambridge, U.K. Yu, H. S. (2000). Cavity expansion methods in geomechanics, Kluwer
Marshall, A. M., Farrell, R., Klar, A., and Mair, R. J. (2012). “Tunnels in Academic, Dordrecht, Netherlands
sands the effect of size, depth, and volume loss on greenfield dis- Yu, H. S., and Houlsby, G. T. (1991). “Finite cavity expansion in dilatant
placements.” Geotechnique, 62(5), 385–399. soils: Loading analysis.” Geotechnique, 41(2), 173–183.

1246 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / OCTOBER 2012


Copyright of Journal of Geotechnical & Geoenvironmental Engineering is the property of American Society of
Civil Engineers and its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email articles for
individual use.

You might also like