See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/239410273

The response of surface structures to tunnel construction

Article in ICE Proceedings Geotechnical Engineering · January 2006

DOI: 10.1680/geng.2006.159.1.3

CITATIONS READS

148 2,196

3 authors, including:

John Boscawen Burland David Potts

Imperial College London Imperial College London
161 PUBLICATIONS 12,148 CITATIONS 336 PUBLICATIONS 10,368 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by John Boscawen Burland on 31 March 2017.

The user has requested enhancement of the downloaded file.

 Proceedings of the Institution of
Civil Engineers
Geotechnical Engineering 158
June 2005 Issue GE1
Pages 1–15

Paper 13751
Received 05/03/2004
Accepted 07/03/2005
Keywords: Jan N. Franzius David M. Potts John B. Burland
mathematical modelling/soil– Geotechnical Engineer, Professor at Imperial College, Professor at Imperial College,
structure interaction/tunnels Geotechnical Consulting London London
Group, London; formerly
Imperial College, London

The response of surface structures to tunnel construction

J. N. Franzius Dipl Ing, PhD, DIC, D. M. Potts BSc, PhD, DSC, FREng, CEng, FICE and J. B. Burland CBE, FRS, FREng,

FS
FICE, FIStructE

Building deformation and the potential damage of

Hogging Sagging
buildings caused by tunnel construction in urban areas
have become a major concern in the planning and
construction process of any tunnelling project. The Building

currently adopted design procedure is highly

inflection
conservative, and can lead to costly projects. In 1997

Point of
Potts and Addenbrooke introduced design charts to
consider the influence of the building’s own stiffness,
thus leading to more realistic predictions of tunnel- Äh
Deflection ratio,
induced deformation. Their relative stiffness approach DR 5 Ä/L Äs
was based on a parametric study using plane-strain
Lh Ls
OO
finite element (FE) analyses in which the building was
modelled by weightless, elastic beams. This paper
presents the results of both two- and three-dimensional
parametric FE studies that extended their building Fig. 1. Definition of deflection ratio
model to include a wider variety of building features
such as building weight, the nature of the soil–
structure interface, and the building dimension in the
direction of the tunnel axis. By incorporating these
additional building features into their approach it is 20·3
shown how the relative stiffness expressions can be
modified to be dimensionless. New design charts that
Ca
Deflection ratio, Ä/L: %

t eg

take account of the wide variety of building

or

20·2
y

characteristics are presented for these new relative

4
an

stiffness expressions.
d
Ca

5
t

da
eg

m
or

ag
y3

1. INTRODUCTION 20·1
e
da

Predicting tunnel-induced deformation of existing structures

m
Ca

ag
PR
te

such as buildings or services, and assessing the risk of potential

e
Ca

g
or
te

damage caused by tunnel construction, are now essential parts

y2
org

0
y1

of the planning, design and construction of tunnels in the

urban environment.1 The current design approach is based on 0 0·1 0·2 0·3
Horizontal strain: %
empiricism gained from greenfield sites, and consequently does
not consider building characteristics such as building stiffness
or building weight. Instead the greenfield deformation Fig. 2. Relationship of damage category to deflection ratio and
horizontal tensile strain for hogging (after Burland4 )
predicted at the location of an existing structure is assessed by
adopting deformation criteria such as deflection ratio (defined
in Fig. 1) and horizontal strain. Interaction diagrams (such as
that shown in Fig. 2), based on elastic deep beam theory,2,3 are In 1997 Potts and Addenbrooke5 proposed the ‘relative stiffness
then used to relate these deformation criteria to categories of approach’ to take the building’s stiffness into account when
potential damage. Aesthetic damage is described by categories predicting its tunnel-induced deformation. This approach was
0, 1 and 2, and categories 3 to 5 affect serviceability and based on over 100 plane-strain finite element (FE) analyses in
structural stability.4 This approach is usually conservative, as it which the building was represented by an elastic beam. In their
assumes that the building is infinitely flexible and follows the parametric study they included a wide range of both bending
greenfield deformation. and axial stiffnesses for different building geometries.

Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al. 1

Article number = 13751

 Franzius6 presented the results of an extensive parametric In their parametric study the building was represented by an
study on this problem, which considered a variety of building elastic beam. The tunnel-induced deformation of this structure
characteristics such as building weight and the nature of the was quantified by its deflection ratio DR and its maximum
soil–building interface. Using 3D FE analysis the effects of horizontal strain
h . These measures were then related to
step-by-step tunnel construction on building deformation were corresponding FE greenfield situations (i.e. no structure

FS
evaluated. In addition the influences on the relative stiffness present) with equivalent geometry by defining the following
approach of different geometric dimensions in both the modification factors:
transverse and longitudinal directions (with respect to the
tunnel) were assessed independently. This paper summarises DRsag
the results from Franzius,6 and shows how the relative stiffness 2a M DRsag ¼
DR GF
sag
method proposed by Potts and Addenbrooke5 can be adjusted
to account for these additional building features, therefore
extending its applicability and providing greater confidence for and
its use in engineering practice.
DR hog
2b M DRhog ¼
2. THE RELATIVE STIFFNESS APPROACH DR GF
hog
Potts and Addenbrooke5 introduced relative stiffness
expressions to relate the stiffness of a surface structure to the
stiffness of the ground. They defined for sagging and hogging deflection ratio, and

hc
r
¼
EI 3a M
hc ¼
1a
GF

and
OO ES ð B=2Þ4

and
hc

ht
EA 3b M
ht ¼
1b a
¼
GF
ht
ES ð B=2Þ

for compressive and tensile maximum horizontal strain

as the relative bending and axial stiffness respectively. EI is the
respectively. The index ‘GF’ denotes the greenfield deformation
bending stiffness of the building, EA the building’s axial
criterion. Note that Potts and Addenbrooke5 adopted the
stiffness, and ES is the soil’s secant stiffness obtained at 0.01%
maximum horizontal strain developing within the geometry of
axial strain in a triaxial compression test performed on a
a structure as a measure, whereas the building damage
sample retrieved from half of the tunnel depth z0 (for the soil
assessment procedure outlined by Mair et al.1 employs the
profile used in their analysis Es was 103 MPa and 163 MPa for
average horizontal strain over the length of the building span.
tunnel depths of z0 ¼ 20 m and 34 m respectively). B is the
width of the building transverse to the tunnel, as defined in
In their work Potts and Addenbrooke5 showed that the
Fig. 3. Potts and Addenbrooke5 applied these relative stiffness
modification factors ar M DRsag and M DRhog obtained from their
expressions to the results of a 2D parametric FE study. In such
parametric study correlated with the relative bending stiffness
a plane-strain situation r* has the dimension [1/length] and Æ*
r*, whereas M
hc and M
ht correlated with the relative axial
is dimensionless.
PR

stiffness Æ*. Upper bounds were then given to these data to

provide a conservative estimate of the variation of the
B modification factors with respect to the structure’s relative
Slab: EC, Islab, Aslab
stiffness. Design charts developed from these curves are shown
Structure: in Fig. 4.
m storeys:
m 1 1 slabs
These curves can be used for design purposes. As a first step
the greenfield settlement is predicted for the tunnel geometry
1x and the expected volume loss VL by adopting empirical
1z expressions (i.e. a Gaussian settlement curve) as described by
London Clay, Es O’Reilly and New.7 Similar expressions are given for the
e
z0 distribution of horizontal strain. For a given building position
with respect to the tunnel axis the greenfield deflection ratio
D
and maximum horizontal strain are calculated. The bending
and axial stiffness of the building must then be evaluated,
leading to the relative stiffness values. With these values the
modification factors can be obtained from the design charts.
Multiplying these factors with the greenfield deformation
Fig. 3. Transverse geometry of the interaction problem criterion leads to modified values that take into account the
building’s stiffness. With these data the building damage

2 Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al.
 1·2

M DRsag
0·8 e/B 5 0 e/B 5 0
0·4 0·2 0·2
0·4 0·08
0·6
0·4

M åhc

FS
0 28 27 26 25 24 23 22 21 0·04
10 10 10 10 10 10 10 10 100 101 r*: m21 0·6
0
0
0·1 1·0 10·0 100·0 á*
0·4 e/B , 0·2 0
M DRhog

0·4
0·8 0·6 0·04 e/B , 0·2

M åht
1·2 0·08 0·4
0·6
(a) (b)

Fig. 4. Design curves for modification factors of (a) deflection ratio and (b) maximum horizontal strain, proposed by Potts and
Addenbrooke5

category can be assessed using interaction diagrams as shown The initial stress profile is prescribed by a saturated bulk unit
in Fig. 2. weight of the soil of ª ¼ 20 kN/m3 and a water table at a depth
of 2 m below the ground surface. The pore water pressure
3. DETAILS OF ANALYSIS

3.1. Geometry
OO
This paper presents results from both 2D and 3D FE analyses.
Fig. 3 shows the geometry adopted for the 2D plane-strain
distribution with depth is hydrostatic, with a zone of suction
above the water table. The coefficient of earth pressure at rest,
K0 ¼ 1.5, was applied over the whole soil profile. This is in
contrast with the study performed by Potts and Addenbrooke,5
in which a zone of reduced K0 was implemented in order to
analyses. The building had a width B and an eccentricity e improve the predictions for the greenfield surface settlement
measured between building and tunnel centre line. In the study trough.9 Such a zone would have been unreasonable in the
presented in this paper the different geometric measures were study presented in this paper, as building features such as
varied independently in order to assess their influence on the building weight were applied to the model before simulating
relative stiffness approach. Potts and Addenbrooke5 included tunnel construction. The reduced-K0 zone would not be
only two tunnel depths z0 in their study (20 m and 34 m), appropriate when simulating the consolidation behaviour
whereas this paper additionally employs z0 ¼ 15 m and 28 m. following the application of building load.
The tunnel diameter was not varied, and a value of D ¼
4.146 m was adopted, as used in the original study by Potts 3.3. Tunnel excavation
and Addenbrooke.5 This value is typical of a running tunnel on In 2D analyses the tunnel construction was simulated using the
the London Underground system. volume loss method, described by Potts and Zdravković.10 The
volume loss VL is defined as the ratio of volume (per metre
In 3D analyses a 100 m-wide building was modelled with its length) of soil moving into the tunnel to the original cross-
building width B perpendicular to the tunnel axis. The building section of the tunnel (D2 /4 per metre length). In undrained
had no eccentricity. The longitudinal dimension of the conditions it can also be determined by measuring the volume
PR

structure was given by the building length L parallel to the (per metre length) of the surface settlement trough and
tunnel axis. dividing this value by the cross-sectional area of the tunnel.

3.2. Soil model and initial stress Two-dimensional tunnel excavation was simulated over 15
The soil model adopted in this study was the same as that increments. At the beginning of this procedure the soil stresses
employed by Potts and Addenbrooke 1 . The whole soil profile that act around the tunnel boundary were evaluated and then
consisted of London Clay, which was represented by a non- applied in the reverse direction over the 15 excavation
linear elasto-plastic constitutive model. The non-linear pre- increments. The elements within the tunnel were not included
yield behaviour was based on the model introduced by Jardine during these increments. The tunnel lining was installed in the
et al.,8 and describes the reduction of G/p9 and K/p9 with shear seventh excavation increment in order to achieve a volume
strain and volumetric strain respectively, by using loss close to VL ¼ 1.5%, which lies well within the range
trigonometric functions. G and K are the tangent shear and observed for tunnelling in stiff overconsolidated clay, such as
bulk modulus respectively. The yield surface and plastic London Clay.11 This value of VL was achieved in the greenfield
potential are described by a Mohr–Coulomb model. The soil analyses, but was not reproduced exactly when buildings were
parameters used for these models are listed in Appendix 1. The included in the analyses. However, the results were adjusted to
results presented in this paper are therefore applicable only to a volume loss of 1.5%, as described by Potts and Addenbrooke5
tunnelling in London Clay or in similar stiff clays. By and Franzius.6
performing a comparison plane-strain analysis it was shown
that modelling soil layers of gravel overlying the London Clay A number of 3D analyses were performed to investigate
does not have a significant effect on the results. different aspects of the tunnel–soil–structure interaction.

Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al. 3
 When the influence of building length L was studied, the 3.5. Finite element program
tunnel was excavated simultaneously over the whole mesh All analyses were carried out using the Imperial College Finite
length, adopting the volume loss method. However, when the Element Program (ICFEP). In plane strain the soil was modelled
influence of the 3D tunnelling sequence on building by 8-node elements, and 3-node beam elements represented the
deformation was investigated, a step-by-step approach12 was building. In 3D analyses 20-node elements were used for the

FS
adopted. For these analyses the 3D mesh was divided in the soil and 8-node shell elements for the structure. Reduced
longitudinal direction into ‘slices’ of elements of equal length integration was used, with an accelerated modified Newton–
Lexc . In one excavation step, elements within the tunnel Raphson scheme with an error controlled substepping stress
boundary were excavated over Lexc . The tunnel lining was point algorithm for solving the non-linear FE equations.10,16
subsequently installed over this distance before the next ‘slice’
of elements was excavated. This sequence was not volume loss A typical mesh used for 3D analyses of step-by-step tunnel
controlled, and the amount of VL is dependent on Lexc . An construction is shown in Fig. 5. Because of symmetry, only one
excavation length of Lexc ¼ 2.5 m was applied during the half of the problem had to be modelled. This particular mesh
study. The results from the 3D analyses were scaled to a had 10 434 solid elements and 46 930 nodes. The plane-strain
volume loss of 1.5% in order to compare 3D and 2D analyses.6 mesh had a similar geometry in the transverse direction. The
The tunnel lining was modelled by elastic shell elements13 2D mesh was, however, slightly finer than the corresponding
representing a continuous concrete lining. transverse section of the 3D mesh shown in Fig. 5.

It was noted by various authors (for example Addenbrooke14 ) 4. INFLUENCE OF GEOMETRY

that modelling tunnel excavation in a high-K0 regime predicts A large number of plane-strain FE analyses were performed in
surface settlement troughs that are too wide when compared order to investigate the influence of the building’s transverse
with field data. Such an effect was also observed in the present geometry and the tunnel depth on the deformation behaviour
OO
study. The relative stiffness approach, however, overcomes this
problem to a significant extent by applying the same FE
simulation to both building and greenfield scenarios and than
of the building. In addition, 3D analyses were performed in
which the building was represented by shell elements having
both a transverse and a longitudinal dimension, referred to as
normalising them against each other. building width B and length L respectively. In the 3D analyses
presented in this section, the tunnel was constructed over the
whole mesh length simultaneously, effectively modelling a
plane-strain excavation beneath a 3D structure. This procedure
3.4. Building model was chosen in order to separate the influence of the 3D
In plane-strain analyses the building was represented by an geometry from other effects caused by the 3D tunnelling
elastic beam with a Young’s modulus E, a Poisson’s ratio , a sequence.
second moment of area I and a cross-sectional area A. In 3D
analyses an elastic shell was used with a Young’s modulus E
4.1. Building width
and thickness t. A building was considered to consist of m
A suite of plane-strain FE analyses was carried out in which
storeys made up of m + 1 concrete slabs, as shown in Fig. 3.
the building width B was varied while the tunnel depth
The second moment of area of the building, I, was calculated
remained constant as z0 ¼ 20 m. No eccentric geometries were
from the second moment of area of each slab, Islab , by
included in this particular study. For all cases modification
employing the parallel axis theorem 15 with the neutral axis at
factors were calculated. Fig. 6 plots the deflection modification
the middle of the building. Axial straining is assumed along
factors M DR against r*. For each building width, four building
each structure’s full height to give the axial stiffness. The beam
stiffnesses, representing one-, three-, five- and ten-storey
was placed on the ground surface and had properties
buildings, were included. The figure also includes the design
PR

equivalent to one-, three-, five- and ten-storey buildings, for

curve proposed by Potts and Addenbrooke.5 The figure shows
which the input parameters are derived in Appendix 2. The
calculation of the building stiffness employing the parallel axis
theorem is likely to overestimate the stiffness of a real
structure. The building foundation, which would also
contribute to the structure’s stiffness, was not modelled. This
omission, in contrast, reduces the overall stiffness of the
buildings modelled in this study. It has to be noted that this
simplified building model does not account for different 20
m
building stiffness along perpendicular directions of the Building
building. It is, however, consistent with the critical strain
x
concept of Burland and Wroth,2 which uses elastic beam theory
z y
to calculate the strain within a structure. 55 m

40 m
In the study by Potts and Addenbrooke5 the building was m
modelled to be weightless and have a rough interface with the
3
1000 m 100
m 30 m
soil. However, the parametric study, of which results are
presented in this paper, also included cases in which the
building weight was modelled and the nature of the contact Fig. 5. Typical 3D finite element mesh
between building and soil was varied.

4 Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al.
 1·6 0·12
Design curve: e/B 5 0 Design curve: e/B 5 0
B 5 16 m B 5 16 m
B 5 32 m 0·10 B 5 32 m
1·2 B 5 60 m B 5 60 m
B 5 100 m 0·08 B 5 100 m
B 5 120 m B 5 120 m

FS
M DRsag

M åhc
0·8 0·06

0·04
0·4
0·02

0 0
1028 1027 1026 1025 1024 1023 1022 1021 100 101 0·1 1·0 10·0 100·0
r*: m21 á*
(a)

r*: m21
Fig. 7. Horizontal compression modification factor against Æ*
28 27 26 25 24 23 22 21 0 1 for buildings of varying width and stiffness, z0 ¼ 20 m
10 10 10 10 10 10 10 10 10 10
0

0·4 4.2. Tunnel depth

A similar suite of analyses as presented in the previous section
M DRhog

0·8 was performed to investigate the influence of the tunnel depth

z0 on the building’s deformation. For a five-storey structure,

1·2

1·6
OOB 5 16 m
B 5 32 m
B 5 60 m
B 5 100 m
B 5 120 m
Design curve: e/b , 0·2

(b)
analyses were carried out with tunnel depths of z0 ¼ 15, 20, 28
and 34 m. Fig. 8 shows the deflection ratio modification factors
obtained from this study. The various symbols refer to different
building geometries. Two eccentric cases were included for
100 m-wide buildings. A vertical pattern can be seen for each
building geometry, with the z0 ¼ 34 m deep tunnel giving the
Fig. 6. Deflection ratio modification factor against r* for highest modification factor, whereas shallow tunnels lead to
buildings of varying width and stiffness, z0 ¼ 20 m: (a) M DRsag ;
(b) M DRhog

1·6

that most cases lie below these design curves, but a significant 1·2
scatter emerges from the M DRsag plot (Fig. 6(a)). The arrow
M DRsag

marks the results for the 120 m, 100 m and 60 m wide one- 0·8
storey buildings. It can be seen that all three cases give similar z0 5 34 m

values of M DRsag , although their relative bending stiffness z0 5 28 m

varies between 1.5 3 105 and 2.4 3 104 m1 . Similar 0·4 z0 5 20 m
patterns can be found for the three-storey buildings. With z0 5 15 m

increasing relative bending stiffness r*, however, the scatter 0

PR

reduces. These results indicate that the influence of B might be 1028 1027 1026 1025 1024 1023 1022 1021 100 101
overestimated in the current definition of r* (equation 1(a)), r*: m21
(a)
where B is raised to the power of 4.
r*: m21
DRhog 1028 1027 1026 1025 1024 1023 1022 1021 100 101
The corresponding M plot (Fig. 6(b)) shows less scatter, 0
and the results for low values of r* follow the design curve.
There are fewer data points in this graph as no hogging was
0·4
obtained for the B ¼ 32 m- and 16 m-wide structures. Only a
small scatter is also evident when plotting the horizontal
B 5 16 m
M DRhog

compression modification factors M
hc against Æ*, as shown in 0·8 B 5 32 m

B 5 60 m
Fig. 7. All results follow the design curve.5 Note that the B 5 100 m
maximum horizontal strain was taken as a measure of the B 5 120 m
1·2 B 5 100m, e 5 20 m
building deformation in all results presented in this paper in B 5 100m, e 5 30 m
order to maintain consistency with the original work of Potts Design curve: e/B , 0·2
Design curve: e/B 5 0·4
and Addenbrooke.5 Tensile strain was observed only in 1·6
(b)
buildings with a certain offset from the tunnel centre line. For
buildings with no eccentricity only compressive horizontal
Fig. 8. Deflection ratio modification factor against r* for five-
strain occurred, despite the buildings showing hogging
storey buildings of different geometries and for tunnels of
deformation (i.e. zones of sagging and compression or of different depth: (a) M DRsag ; (b) M DRhog
hogging and tension did not coincide in the building cases).

Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al. 5
 lower factors. This trend is most obvious for the M DRsag of vertical mesh boundary (30 m) was chosen such that the
buildings with a low relative stiffness, where results for greenfield settlement trough that develops at the remote
different tunnel depths are labelled in Fig. 8(a) for the 120 m- boundary is equal to the settlement profile obtained from a
wide building. The vertical pattern indicates that the influence greenfield plane-strain analysis.
of the tunnel depth is underestimated by the relative bending

FS
stiffness formulation (equation 1(a)). In this equation the tunnel A parametric study was performed for L ¼ 1, 2, 4 and 8 m. For
depth is represented only by the soil stiffness taken at half all geometries one-, three-, five- and ten-storey structures were
tunnel depth. Fig. 8 indicates that the direct incorporation of z0 considered. When calculating the relative stiffness expressions
would be more appropriate. it was possible to consider L in this calculation by including it
in the second moment of area I and in the area A (rather than
Figure 9 presents similar graphs for strain modification factors calculating them per running metre). This leads to 3D relative
plotted against Æ*. For compressive strain (Fig. 9(a)) all data
and Æ
, where r
is dimensionless
stiffness expressions r3D 3D 3D
points follow the pattern of the design curve (given for no

and Æ3D has the dimension [length].
eccentricity). Tensile strain (Fig. 9(b)) was obtained only in
analyses with eccentric buildings. It can be seen that the results Figure 10 plots the deflection ratio modification factors from
for tensile strain follow the design curves, although the results the 3D analyses against the corresponding r
3D . In addition the
for eccentricity e ¼ 30 m (e/B ¼ 0.3) lie slightly above the e/B graphs also include modification factors obtained from
¼ 0.4 design curve. These graphs indicate that an indirect corresponding plane-strain analyses for each building stiffness.
representation of z0 in the relative axial stiffness formulation These results are plotted against the plain strain relative
(z0 is represented in the soil stiffness taken at half tunnel bending stiffness r*. Note that although the L ¼ 1 m scenarios
depth in equation 1(b)) leads to a good correlation between and the corresponding plane-strain cases have the same
M
h and Æ*.
and r* respectively), the
magnitude of relative stiffness (r3D

4.3. Building length

OO
As noted above, the influence of the building length L,
dimensions of the two relative stiffness measures are different.
The graph also shows the design curves, derived by Potts and
Addenbrooke5 from plane-strain results.
longitudinal to the tunnel axis, was studied by performing a
suite of 3D analyses in which the tunnel was constructed For each group of 3D results the L ¼ 1 m case gives the highest
simultaneously over the whole mesh length. Because of modification factor, and the 8 m-long building leads to a
symmetry only one quarter of the problem had to be modelled. modification factor of similar magnitude as for the
The longitudinal distance between building and the remote

1·6
Design curve (plane strain)
0·12 1 storey
Design curve: e/B 5 0
B 5 16 m 3 storeys
0·10 B 5 32 m 1·2 5 storeys
B 5 60 m 10 storeys

0·08 B 5 100 m
M DRsag

B 5 120 m
B 5 100 m, e 5 20 m 0·8
M åhc

0·06 B 5 100 m, e 5 30 m

0·04 0·4

0·02
0
0 1028 1027 1026 1025 1024 1023 1022 1021 100 101
PR

0·1 1·0 10·0 100·0 r*3D; r*: m21

á* (a)
(a)
r*3D; r*: m21
28 27 26 25
á* 10 10 10 10 1024 1023 1022 1021 100 101
0
0·1 1·0 10·0 100·0
0

0·02 0·4

0·04
M DRhog

0·8
M åht

0·06
1 storey
0·08 1·2 3 storeys
Design curve: e/b , 0·2 5 storeys
Design curve: e/b 5 0·4 10 storeys
0·10 Design curve (plane strain)
B 5 100 m, e 5 20 m
B 5 100 m, e 5 30 m 1·6
0·12 (b)
(b)
Fig. 10. Deflection ratio modification factors from 3D analyses
Fig. 9. Horizontal strain modification factor against Æ* for five-
calculated for 3D geometry
(solid symbols) against r3D
storey buildings of different geometries and for tunnels of together with plane-strain results (hollow symbols) against r*:
different depth: (a) M
hc ; (b) M
ht (a) M DRsag ; (b) M DRhog

6 Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al.
corresponding plane-strain result. However, the variation of
0·12
modification factor over L is relatively small for all sagging Design curve (plane strain)
1 storey
cases and for most hogging cases when compared with the 0·10 3 storeys
vertical scatter obtained for different tunnel depths (Fig. 8). 5 storeys
0·08 10 storeys
This trend becomes clearer when all modification factors are

FS
plotted against the corresponding plane-strain relative stiffness

M åhc
0·06
r*. These graphs are shown in Fig. 11. All results, apart from
the hogging factors for the three-storey structures, lie within 0·04
the boundaries described by the Potts and Addenbrooke design
curves.5 These graphs demonstrate that the description of 0·02

relative bending stiffness by using plane-strain measures for

0
3D structures provides a good estimate for predicting MDR . 0·1 1·0 10·0 100·0
á*3D: m; á*

Figures 12 and 13 present the corresponding graphs for

compressive strain modification factors. It can be seen that the Fig. 12. Horizontal compression modification factors from 3D

calculated for 3D
analyses (solid symbols) against Æ3D
change in ar M
hc with L is much more significant than for
geometry, together with plane-strain results (hollow symbols)
MDR . When plotting the 3D results against the plane strain r* against Æ*
(Fig. 13) the results lie outside the Potts and Addenbrooke
design curves5 , whereas the corresponding plane-strain
modification factors lie just below them.

0·12
It is interesting to note that horizontal building strain (and Design curve
OO
therefore horizontal building deformation) is affected much
more by changes in building length L than the deflection ratio
(calculated from the vertical displacement) is. The increases of
M
hc in Fig. 13 appear to be significant. However, it should be
noted that the strain modification factors are still small in
M åhc
0·10

0·08

0·06
1 storey
3 storeys
5 storeys
10 storeys

magnitude. Furthermore, building lengths of L ¼ 4 m and 0·04

smaller are an unrealistic geometry when considering entire
0·02

0
1·6 0·1 1·0 10·0 100·0
Design curve á*
1 storey
3 storeys
1·2 5 storeys Fig. 13. Horizontal compression modification factors from 3D
10 storeys (solid symbols) and 2D (hollow symbols) analyses against Æ*
(calculated for plane strain)
M DRsag

0·8

0·4 buildings of relatively large width (B ¼ 100 m), and were

included only as extreme values of L.
0
PR

1028 1027 1026 1025 1024 1023 1022 1021 100 101 5. FURTHER BUILDING AND TUNNEL
r*: m21
CHARACTERISTICS
(a)
Apart from the building geometry, further building features
r*: m21 were included in the parametric study, and their influence on
1028 1027 1026 1025 1024 1023 1022 1021 100 101 the tunnel-induced building deformation was assessed. The
0
findings from these studies are briefly summarised in the
following paragraphs.
0·4

5.1. Building weight

M DRhog

0·8 A parametric study was performed in which a wide range of

building stiffness–load combinations was considered (varying
1 storey between 10 and 100 kPa). The building load was applied at the
1·2 3 storeys
5 storeys beginning of the FE analysis, and the soil was allowed to
10 storeys consolidate fully under the load before tunnel excavation
Design curve
1·6 commenced. Further details of these analyses can be found in
(b)
Franzius et al.17 It was found that generally the magnitudes of
both MDR and M
h increased with building load. However,
Fig. 11. Deflection ratio modification factors from 3D (solid
when realistic stiffness–load combinations (10 kPa per storey)
symbols) and 2D (hollow symbols) analyses against r*
(calculated for plane strain): (a) M DRsag ; (b) M DRhog were modelled, the rapid decrease of modification factors with
increasing building stiffness (as observed also for no-weight

Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al. 7
 cases) compensated for the slight increase of modification relative bending stiffness expression are not appropriately
factors due to building load. It was concluded that the Potts represented.
and Addenbrooke design curves5 were still appropriate. (c) Potts and Addenbrooke5 modelled the building in plane
strain, having no self-weight and a rough interface with
5.2. Soil–building interface the soil. The parametric study summarised above studied

FS
Interface elements18 were included in plane-strain analyses to the influence of these building characteristics, and in
vary the nature of the contact between soil and building addition modelled 3D tunnel construction.
foundation. When potential separation between building and
soil was allowed, it was shown that it did not develop as long This section will describe how the relative stiffness expressions
as building weight was considered. When horizontal relative can be modified to take into account the above shortcomings.
movement between soil and building was allowed (i.e. smooth
interface), the horizontal strain in the building vanished. 6.1. Relative bending stiffness
Compared with the large reduction in M
h values, MDR values It was shown previously that the influence of building width B
were less affected, showing a small reduction with decreasing in the original definition of r* (equation 1(a)) was
interface shear stiffness (see Franzius6 for further details). overestimated and the tunnel depth z0 was not sufficiently
Again it was concluded that the design curves by Potts and weighted. It was furthermore demonstrated that the variation in
Addenbrooke5 were appropriate. MDR with varying building length L is small. Defining

5.3. 3D tunnel construction process EI

¼
rmod
In a suite of 3D FE analyses the tunnel construction was 1
ES z0 B2 L
modelled by adopting a step-by-step construction sequence.
Buildings of different geometry and stiffness were included in
OO
the study. In plan view all buildings were perpendicular to the
tunnel axis, with B (being the longer dimension) in the
transverse and L in the longitudinal direction. No tensile
reduces the influence of B compared with the original
formulation while z0 is incorporated into the relative stiffness
expression. The soil stiffness in the above equation is defined
horizontal strain was obtained in any of the non-eccentric in the same way as in equation 1(a), i.e. taken at z0 /2. For the
structures. Building deformation in the transverse direction to building’s second moment of area I, the 3D expression, having
the tunnel was compared with the results from plane-strain the dimension [length4 ], is applied. Dividing by L, however,
analyses. Generally it was found that deformation criteria and always reduces it to its plane-strain expression. Alternatively,
corresponding modification factors obtained from analyses of one can use the plane-strain value of I (per running metre) and
3D tunnel excavation were in good agreement with results divide by a unit length of L ¼ 1 m/m (i.e. per running metre). It
from corresponding plane-strain analyses when situations can be seen that the above equation always remains
involving a similar volume loss were compared.6 dimensionless. The use of B in the above equation rather than
B/2 as in the original formulation has no implication for the
Hogging and horizontal strain developed within the building in relative position of the results to each other when plotting
the longitudinal direction as the tunnel approached and passed against a log-scale of relative stiffness. The use of B is more
under the building. For the building modelled in this study this consistent with expressing the degree of eccentricity as e/B.
deformation was, however, smaller in magnitude than that
observed in the transverse direction,6 and it is therefore not Figure 14 presents the modification factors M DR for buildings
further discussed in this paper. with a wide range of different geometries, and plots them
against r
. All results in this figure are from plane-strain
mod
6. ADJUSTMENT OF RELATIVE STIFFNESS analyses with weightless buildings with a rough interface
PR

EXPRESSIONS between building and soil but with varying building stiffness
The relative stiffness method has been described as the most and width and tunnel depth (i.e. building characteristics similar
promising development of recent years for taking into account to those modelled by Potts and Addenbrooke5 ). Only
the relative stiffness between soil and structure,19 and it has geometries with no eccentricity are plotted in the figure. For
been applied for settlement predictions in London.20 The both M DRsag and M DRhog there is a well-defined upper bound.
parametric study summarised in the previous sections, The upper-bound curves shown in Figs 14(a) and (b) are similar
however, highlighted the following shortcomings in the in shape to those defined by Potts and Addenbrooke.5
relative stiffness formulation: However, by plotting the same results against their original
definition of relative stiffness r*, Franzius6 demonstrated that
(a) In plane strain r* has the dimension [1/length] and Æ* is for M DRsag the scatter in the data reduces significantly, whereas
dimensionless. This changes under 3D conditions, where r* for M DRhog it is only marginally changed.
becomes dimensionless and Æ* has the dimension [length].
It would be more consistent if both r* and Æ* were Figure 14 includes only the building cases that adopted the
dimensionless regardless of whether applied in 2D or in 3D same simplifications as used by Potts and Addenbrooke.5 Fig.
situations. 15 plots MDR from analyses for cases either with building
(b) Including a wider variation of building geometry and weight, different soil–foundation properties, varying building
tunnel depth than investigated by Potts and Addenbrooke5
. Only
length, or subjected to 3D tunnel excavation against rmod
revealed a scatter when plotting modification factors results from 100 m-wide structures with no eccentricity but
against the corresponding relative stiffness values. It was varying building stiffness and tunnel depth are presented. The
shown that the building width B and tunnel depth z0 in the various symbols focus on different building features: ‘Weight’

8 Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al.
 1·6 1·6
Upper bound

1·2 1·2

FS
M DRsag

M DRsag
0·8 0·8

0·4 0·4

0 0
1026 1025 1024 1023 1022 1021 100 101 102 103 1026 1025 1024 1023 1022 1021 100 101 102 103
r*mod r*mod
(a) (a)
r*mod r*mod
1026 1025 1024 1023 1022 1021 100 101 102 103 1026 1025 1024 1023 1022 1021 100 101 102 103
0 0

0·4 0·4
M DRhog

M DRhog
0·8 0·8 Standard
Weight

1·2

1·6
OO Upper bound

(b)
1·2

1·6
(b)
L51
L54
IF

3D, 20
3D, 30
Upper bound e/B 5 0

Fig. 14. Deflection ratio modification factors for cases with no Fig. 15. Deflection ratio modification factors for cases with no
eccentricity plotted against modified relative bending stiffness eccentricity but additional building features plotted against
r
mod
: (a) M DRsag ; (b) M DRhog
modified relative bending stiffness rmod

and ‘IF’ refer to analyses modelling realistic building weight modification factors for buildings with eccentricity are
and a soil–structure interface with low interface shear-stiffness presented in Fig. 16. Only ‘standard’ building scenarios are
respectively. The 3D structures with an out-of-plane dimension included in these figures. Separate upper-bound curves are
of L ¼ 1 m and 4 m, subjected to plane-strain tunnel added to the data for different ranges of eccentricity (defined
construction, are assigned by the labels ‘L ¼ 1’ and by e/B). It can be seen that for M DRsag an increase in
‘L ¼ 4’ respectively, and ‘3D, 20’ and ‘3D, 30’ represent data eccentricity leads to lower modification factors, whereas for
from analyses modelling fully 3D tunnel construction with hogging the opposite is the case.
building lengths of L ¼ 20 m and 30 m respectively. Buildings
with no additional features but with the same geometry as 6.2. Relative axial stiffness
PR

these special cases (i.e. which were included in the previous It was highlighted in this paper that the dimension of Æ* is
plots) are referred to as ‘Standard’. The upper-bound curves [length] when analysing 3D structures, compared with a
fitted to the data in Fig. 14 are also shown in the graphs of Fig. dimensionless expression in plane-strain situations. To
15. When comparing results with the same value of rmod
(i.e. overcome this problem the building length L can be included in
with the same transverse geometry) the graph shows increases the relative axial stiffness definition as
in MDR (compared with the ‘standard’ analyses) for analyses
including weight and a small value of building length L. Most EA
of the analyses with fully 3D tunnel excavation show a 2 Æ
¼
ES BL
moderate reduction in MDR , whereas modelling an extremely
low interface shear stiffness causes a significant decrease in
MDR . All results lie inside the upper-bound curves despite the
it can be seen that Æ

Similar to the definition of rmod mod
wide variation of different building parameters. remains dimensionless whether adopted in 2D or 3D. Equation
(5) is similar to the original definition of Æ* (equation 1(b))
It can be seen that some modification factors exceed unity. apart from the half building width (B/2) being replaced by B,
This behaviour is due to the changed surface boundary which shifts the results to the left when plotted against a
conditions between greenfield and building situations, and has horizontal log-scale axis.
been discussed by Potts and Addenbrooke.5
In Fig. 17 M
h for both eccentric and concentric ‘standard’
The previous figures focused only on building geometries with building cases are plotted against Æ
mod . New upper-bound
no eccentricity with respect to the tunnel axis. Deflection ratio curves are fitted to the data. As with the original curves

Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al. 9
 stiffness are not represented in this figure, as the tunnel-
1·6
Upper bounds induced strain in the building was so small that the
e/B 5 0 corresponding modification factors were more than two orders
e/B 5 0·2
1·2 e/B 5 0·4 of magnitude smaller than M
h obtained from the
e/B 5 0·6 corresponding rough-interface analyses.

FS
M DRsag

0·8
It can be seen that building weight slightly increases the
modification factors for both tension and compression (as
0·4 discussed in Franzius et al.17 ). Fully 3D tunnel excavation (‘3D,
20’and ‘3D, 30’) also has a small influence, with 20 m-long
buildings tending to increase M
hc whereas 30m long structures
0
1026 1025 1024 1023 1022 1021 100 101 102 103 reduce its value. The largest increases, however, are due to
r*mod small values in building length L, as discussed previously. The
(a)
graph shows that data points for buildings with L ¼ 1 m lie
r*mod outside the upper bounds that were fitted to the data shown in
1026 1025 1024 1023 1022 1021 100 101 102 103 Fig. 17. However, the length of L ¼ 1 m was chosen as an
0
extreme case and is unrealistic when the stiffness of an entire
Upper bounds building is considered. Results for L ¼ 4 m, which is still a
0·4 e/B 5 0 small value for building length, are also included in the graph.
e/B 5 0·2
e/B 5 0·4 It can be seen that the changes in M
hc imposed by these
M DRhog

0·8
e/B 5 0·6 building cases are much smaller than for L ¼ 1 m. M
hc for
4 m-long buildings lie inside the upper-bound curve. The

1·2
OO e/B , 0·2
e/B 5 0·2
0·2 , e/B , 0·4
e/B 5 0·4
0·4 , e/B , 0·6
e/B > 0·6
results for L ¼ 8 m lie further inside the upper bounds and are
not included in the graph.

1·6 7. APPLICATION
(b)
In their initial work Potts and Addenbrooke5 described how the
Fig. 16. Deflection ratio modification factors for cases with relative stiffness approach could be incorporated into the
eccentricity plotted against modified relative bending stiffness three-stage methodology used for the Jubilee Line Extension
r
mod : (a) M DRsag ; (b) M DRhog Project, London.1 They suggested that their approach can be
incorporated into the second stage assessment, in which the

proposed by Potts and

0·15
Addenbrooke,5 the values of Upper bounds: e/B 5 0
e/B 5 0·2
M
h are small in magnitude e/B 5 0·4
compared with the e/B 5 0·6
Results from parametric studies: e/B , 0·2
corresponding upper-bound 0·10 e/B 5 0·2
0·2 , e/B , 0·4
curves of the MDR graphs. e/B 5 0·4
M åhc

0·4 , e/B , 0·6

e/B > 0·6

The results of analyses in 0·05

PR

which additional building

features were considered are
shown in Fig. 18. For the 3D 0
0·1 1·0 10·0 100·0
studies (labelled as ‘L ¼ 1’; á*mod
‘L ¼ 4’; ‘3D, 20’ and ‘3D, 30’) (a)
only non-eccentric building á*mod
cases were analysed. No 0·1 1·0 10·0 100·0
0
tensile horizontal building
strain was obtained in these
analyses. Data points for
0·05
plane-strain analyses
M åht

considering building weight Upper bounds: e/B , 0·2

also include eccentric e/B 5 0·4
e/B 5 0·6
0·10 Results from parametric studies: e/B , 0·2
building scenarios (e ¼ 20 m, e/B 5 0·2
B ¼ 100 m, z0 ¼ 20 m and 0·2 , e/B , 0·4
e/B 5 0·4
34 m) in which tensile strain 0·4 , e/B , 0·6
e/B > 0·6
developed in the building 0·15
(b)
during tunnel construction.
Fig. 17. Horizontal strain modification factors plotted against modified relative axial stiffness
Analyses with interface Æ
mod : (a) M
hc ; (b) M
ht
elements of low shear

10 Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al.
 building characteristics, the
0·15
Upper bounds: e/B 5 0·0 modification factors M
h
Standard
Weight remain very small. Similar
L51
L54 conclusions were drawn from
0·10 3D, 20
3D, 30 measured building
Standard, e/B 5 0·2

FS
M åhc

Weight, e/B 5 0·2 deformation along the Jubilee

Line Extension.19 The study
0·05 presented here confirms this
decrease, leading to further
confidence in justifying a
0 reduction in horizontal strain
0·1 1·0 10·0 100·0
á*mod when estimating the tunnel-
(a) induced building damage
á*mod along a tunnel project in an
0·1 1·0 10·0 100·0 urban area.
0

8. CONCLUSIONS
0·05
This paper summarises the
modification factors obtained
M åht

from a wide range of

different building scenarios,
0·10
and correlates this data with

0·15
OO (b)
Upper bounds: e/B , 0·2
relative building stiffness.
This approach is based on the
relative stiffness method
proposed by Potts and
Addenbrooke.5
Fig. 18. Horizontal strain modification factors for cases with additional building features plotted

: (a) ar M
hc ; (b) M
ht
against modified relative axial stiffness Æmod
In their original study5 Potts
and Addenbrooke included a
building strain is evaluated for each individual building within wide range of different geometries, including two tunnel
the 10 mm settlement contour (and/or the 1/500 slope of the depths. This paper presented a larger number of analyses in
settlement trough). The standard damage assessment procedure which additional building features, such as building weight, a
requires the building strain to be estimated from the greenfield smooth interface between the soil and the foundation, and the
ground movement. Potts and Addenbrooke5 proposed to adopt longitudinal dimension in the direction of the tunnel axis, were
their modification factors to refine the assessment of the modelled. Additional tunnel depths were also included. These
building settlement prior to calculate the building strain. analyses showed that their design concept is still applicable
when data with these additional building characteristics are
The individual assessment of single buildings along a large- considered.
scale tunnelling project is very time consuming, and a more
efficient approach is currently under development for the It was, however, shown that redefining the relative bending
Crossrail project in London.21 In this methodology the building stiffness expressions r* can reduce the scatter observed when
PR

strain is estimated generically from the surface ground plotting modification factors against r*. Furthermore, it was
movement for a wide range of possible building geometries demonstrated that including the building length (parallel to the
along a tunnel line. Only areas in which this assessment tunnel axis) in the relative stiffness definition leads to a
highlights a potential building damage (exceeding damage dimensionless expression, in contrast to the original
category 2 as given by Burland4 ) does an individual building formulation. New upper-bound curves, shown in Fig. 19, were
assessment have to be performed. In this stage the findings fitted to the deflection modification factors when plotted
presented in this paper could be incorporated into the building against r
mod .
damage assessment.
For axial strain it was found that the original definition of
It is clear that the parametric study presented here is based on relative axial stiffness Æ* gives a good correlation with M
h
certain assumptions regarding the building type or the for all cases analysed. The expression was, however,
situation of a single tunnel. Engineering judgement is therefore rearranged to be dimensionless when applied to both 2D and
required when adapting the results from this study to buildings 3D conditions. As for the relative bending stiffness, the half
with sensitive architectural features or to more complicated building width used by Potts and Addenbrooke5 was replaced
tunnel situations. Furthermore, the estimation of the overall by B to be more consistent with the degree of eccentricity,
stiffness of a structure can be difficult. However, the results which is expressed as e/B. For this modified relative axial
presented here can be used as a design guideline. Of particular stiffness expression new upper-bound curves, shown in Fig.
significance is the vast reduction in horizontal building strain 20, were fitted to the data. These curves can then be
when the structure’s stiffness is taken into account. This paper incorporated into a building damage assessment as outlined
has demonstrated that, even when considering a wide range of by Potts and Addenbrooke.5

Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al. 11
 1·6
New design curve: e/B 5 0
New design curve: e/B 5 0·2
New design curve: e/B 5 0·4
New design curve: e/B 5 0·6
1·2

FS
M DRsag 0·8

0·4

0
1027 1026 1025 1024 1023 1022 1021 100 101 102
r*mod 5 EI/ESz0B 2L
(a)
r*mod 5 EI/ESz0B 2L
27 26 25 24
10 10 10 10 1023 1022 1021 100 101 102
0

0·4
OO
M DRhog

0·8

1·2
New design curve: e/B 5 0
New design curve: e/B 5 0·2
New design curve: e/B 5 0·4
New design curve: e/B 5 0·6
1·6
(b)

: (a) M DRsag ; (b) M DRhog

Fig. 19. Proposed design curves for MDR adopting the modified relative bending stiffness rmod

This paper demonstrated that the relative stiffness approach for

G BªX ª1
predicting tunnel-induced building subsidence can be adopted ¼ A þ B cosðX ª Þ  sinðX ª Þ with
p9 2:303
for a wide range of building scenarios. Incorporating the 6  
building length into the relative stiffness formulations yields Ed
X ¼ log10 pffiffiffiffiffiffi
dimensionless expressions. The new upper-bound curves 3C
presented here provide design guidance when assessing tunnel-
induced building deformation.
PR

K9 SY 1
¼ R þ S cosðY  Þ  sinðY  Þ with
p9 2:303
9. ACKNOWLEDGEMENTS 7  
This study was part of a research project funded by the j
v j
Y ¼ log10
Engineering and Physical Sciences Research Council (EPSRC) T
with industrial collaboration with the Construction Industry
and Research Information Association (CIRIA), the
Geotechnical Consulting Group (GCG), and London
where G and K9 are the tangent shear and bulk moduli
Underground Limited (LUL)
respectively, Ed is the deviatoric strain invariant, and
v is the
volumetric strain. A, B, C, R, S, T, , , ª, and  are constants
whose values are listed in Table 1. Ed,max, Ed,min ,
v,max and
APPENDIX 1: SOIL MODEL
v,min define strain limits above or below which the stiffness
The pre-yield behaviour of the soil is described by a non-linear varies only with p9 and not with strain. Minimum values of
constitutive model.8 It uses trigonometric expressions to model tangent shear and bulk moduli are given by Gmin and K9min
the change in secant soil stiffness with strain level. To respectively.
incorporate this model in a finite element analysis, the secant
stiffness expressions are differentiated and then normalised In addition a Mohr–Coulomb model is employed to simulate
against mean effective stress p9, leading to the following post-yield behaviour in this study. The strength parameters are
tangent values:16 c9 ¼ 5.0 kPa, 9 ¼ 258 and an angle of dilation ł ¼ 12.58.

12 Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al.
 0·20
New design curve: e/B 5 0
New design curve: e/B 5 0·2
New design curve: e/B 5 0·4
New design curve: e/B 5 0·6

0·15

FS
M åhc
0·10

0·05

0
0·1 1·0 10·0 100·0
á*mod 5 EA/ESBL
(a)

á*mod 5 EA/ESBL
0·1 1·0 10·0 100·0
0

0·05
OO
M åht

0·10

0·15

New design curve: e/B , 0·2

New design curve: e/B 5 0·4
New design curve: e/B 5 0·6
0·20
(b)

: (a) M
hc ; (b) M
ht

Fig. 20. Proposed design curves for M
h adopting the modified relative axial stiffness Æmod

Shear modulus
A B C: %  ª Ed,min : % Ed,max : % Gmin : kPa

373.3 338.7 1.0 3 104 1.335 0.617 8.660 25 3 104 0.692 82 2333.3
PR

Bulk modulus
R S T: %  
v,min : %
v,max : % Kmin : kPa

549.0 506.0 1.0 3 103 2.069 0.42 5 3 103 0.15 3000.0

Table 1. Input parameters for isotropic pre-yield model

APPENDIX 2: BUILDING MODEL The axial and bending stiffness of the structure were then
A building with m storeys consists of m + 1 slabs with a calculated as:
vertical spacing of 3.4 m (see Fig. 3). Each slab has the
following properties: 8 ð EC AÞstruct ¼ ð m þ 1Þð EC AÞslab

Young’s modulus, EC ¼ 23.06 kN/m2

Thickness, t ¼ 0.15 m
X
mþ1 
9 ð EC I Þstruct ¼ EC I slab þ Aslab H 2m
leading to the following plane-strain properties: 1

Area, Aslab ¼ 0.150 m2 /m

Second moment of area, Islab ¼ 2.81254 m4 /m where Hm is the vertical distance between the structure’s and

Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al. 13
 Building Bending stiffness, Axial stiffness, Conference on Geotechnical Aspects of Underground
EI: kNm2 /m EA: kN/m Construction in Soft Ground, London, 1996, pp. 713–718.
2. BURLAND J. B. and WROTH C. P. Settlement of buildings and
Slab 6.47 3 103 3.45 3 106 associated damage. Proceedings of the Conference on
One-storey 2.00 3 107 6.90 3 106 Settlement of Structures, Cambridge, 1974, pp. 611–654.
Three-storey 2.00 3 108 1.38 3 107

FS
3. BOSCARDIN M. D. and CORDING E. J. Building response to
Five-storey 6.96 3 108 2.07 3 107
Ten-storey 4.39 3 109 3.80 3 107 excavation-induced settlement. Journal of Geotechnical
Engineering, ASCE, 1989, 115, No. 1, 1–21.
Poisson’s ratio  ¼ 0.15. 4. BURLAND J. B. Assessment of risk of damage to buildings
due to tunnelling and excavation: Invited special lecture.
Table 2. Stiffness of buildings (per unit length). An m-storey Proceedings of the 1st International Conference on
building consists of m + 1 slabs Earthquake Geotechnical Engineering, Tokyo.
5. POTTS D. M. and ADDENBROOKE T. I. A structure’s influence
on tunnelling-induced ground movements. Proceedings of
the Institution of Civil Engineers—Geotechnical Engineering,
1997, 125, 109–125.
the slab’s neutral axis. The bending and axial stiffness values 6. FRANZIUS J. N. The Behaviour of Buildings due to Tunnel
for buildings of different number of storeys are given in Table Induced Subsidence. PhD thesis, Imperial College,
2. In order to get a consistent set of parameters, the above University of London, 2004.
stiffness expressions have to be converted to 7. O’REILLY M. P. and NEW B. M. Settlements above tunnels in
the United Kingdom: their magnitude and prediction.
Proceedings of Tunnelling ’82, London, pp. 55–64.
Efe t 3fe L
10
OO
ð EC I Þstruct ¼

ð EC AÞstruct
12
¼ Efe t fe L
8. JARDINE R. J., POTTS D. M., FOURIE A. B. and BURLAND J. B.
Studies of the influence of non-linear stress–strain
characteristics in soil–structure interaction. Géotechnique,
1986, 36, No. 3, 377–396.
where tfe is the thickness of the beam or the shell used in the 9. ADDENBROOKE T. I. Numerical Analysis of Tunnelling in Stiff
FE analysis and L is the out-of-plane dimension of the Clay. PhD thesis, Imperial College, University of London,
structure, longitudinal to the tunnel axis. Rearranging 1996.
equations (10) leads to 10. POTTS D. M. and ZDRAVKOVIĆ L. Finite Element Analysis in
Geotechnical Engineering: Application, Vol. 1. Thomas
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Telford, London, 2001.
12ð EC I Þstruct 11. MACKLIN S. R. The prediction of volume loss due to
11a t fe ¼
ð EC AÞstruct tunnelling in overconsolidated clay based on heading
geometry and stability number. Ground Engineering, 1999,
32, No. 4, 31–34.
and 12. KATZENBACH R. and BRETH H. Nonlinear 3D analysis for
NATM in Frankfurt Clay. Proceedings of the 10th
ð EC AÞstruct International Conference on Soil Mechanics and
11b Efe ¼
t fe Foundation Engineering, Stockholm, 1981, 1, 315–318.
13. SCHROEDER F. C. The Influence of Bored Piles on Existing
Tunnels. PhD thesis, Imperial College, University of
PR

London, 2003.
In 2D analyses the area and the second moment of area is
14. ADDENBROOKE T. I., POTTS D. M. and PUZRIN A. M. The
calculated per unit length as
influence of pre-failure soil stiffness on the numerical
analysis of tunnel construction. Géotechnique, 1997, 47,
12a Afe ¼ t fe No. 3, 693–712.
15. TIMOSHENKO S. (1955). Strength of Materials, Part 1:
Elementary Theory and Problems, 3rd edn. D. Van
and Nostrand, New York.
16. POTTS D. M. and ZDRAVKOVIĆ L. Finite Element Analysis in
t3 Geotechnical Engineering: Theory, Vol. 1. Thomas Telford,
12b I fe ¼ fe
12 London, 1999.
17. FRANZIUS J. N., POTTS D. M., ADDENBROOKE T. I. and BURLAND
J. B. The influence of building weight on tunnelling-
whereas in 3D analyses the thickness tfe is used directly as an induced ground and building deformation. Soils and
input parameter instead. Foundations, 2004, 44, No. 1, 25–38.
18. DAY R. A. and POTTS D. M. Zero thickness interface
REFERENCES elements: numerical stability and application. International
1. MAIR R. J., TAYLOR R. N. and BURLAND J. B. Prediction of Journal for Numerical and Analytical Methods in
ground movements and assessment of risk of building Geomechanics, 1994, 18, 689–708.
damage due to bored tunnelling. Proceedings of the 19. BURLAND J. B., STANDING J. R. and JARDINE F. M. Assessing

14 Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al.
 the risk of building damage due to tunnelling: lessons from Publication 200, Thomas Telford, London, 2001, Vol. 1,
the Jubilee Line Extension, London. Proceedings of the Ch. 14.
Conference on Geotechnical Engineering: Meeting Society’s 21. HARRIS D. I. and FRANZIUS J. N. Settlement assessment for
Need, Hong Kong, 2001, Vol. 1, pp. 17–44. running tunnels: a generic approach. International
20. MAIR R. J. and TAYLOR R. N. Elizabeth House: settlement Symposium on Geotechnical Aspects of Underground

FS
predictions. In Building Response to Tunnelling (BURLAND Construction in Soft Ground, Amsterdam, 2005
J. B., STANDING J. R. and JARDINE F. M. (eds)), CIRIA Special (submitted).

What do you think?

To comment on this paper, please email up to 500 words to the editor at [email protected]
Proceedings journals rely entirely on contributions sent in by civil engineers and related professionals, academics and students. Papers
should be 2–5000 words long, with adequate illustrations and references. Please visit www.thomastelford.com/journals for author
guidelines and further details.

OO
PR

Geotechnical Engineering 158 Issue GE1 The response of surface structures to tunnel construction Franzius et al. 15

View publication stats