Analysis of jacking forces

during microtunnelling in limestone

Dr. Marco Barla
Research Associate, Department of Structural and Geotechnical Engineering, Politecnico di Torino,
corso Duca degli Abruzzi 24, 10129 Torino, Italy

Mr. Marco Camusso

Student, Department of Structural and Geotechnical Engineering, Politecnico di Torino,
corso Duca degli Abruzzi 24, 10129 Torino, Italy

Mrs. Santina Aiassa

Project Engineer, Desa srl, via Buozzi 2, 10124 Torino, Italy

Keywords: jacking forces, numerical analyses, microtunnelling, limestone

Abstract
This paper is intended to describe an Italian case study of a microtunnelling project where the boring
machine got stuck during jacking of a 760 mm pipe in a limestone formation. Reference is made to rock
mass characterisation, including site investigations and laboratory tests. Machine’s performance is com-
pared to prediction. To allow for a correct understanding of the conditions which led to the rising of unex-
pectedly high jacking forces, continuum and discontinuum numerical analyses are used. The latter are
shown to be essential at the design analysis stage, when dealing with microtunnelling in a rock mass.

1. Introduction
Microtunnelling operations allow the installation of pipelines from jacking pits, thus minimising the pollu-
tion due to pipe laying in trenches. During the microtunnelling process, the jacking load is an important
parameter, controlling the pipe wall thickness, the need for and location of intermediate jacking stations,
selection of jacking frame, and lubrication requirements. The main component of the jacking load is due
to frictional resistance.
Microtunneling is a complex process and this makes the theoretical determination of these forces quite
difficult. The pipe-soil/rock interaction is in fact highly affected by excavation conditions (overcut, lubrica-
tion, pipe misalignment, stoppages, tunnel stability, rock mass quality, etc.). Moreover, although many
studies have been performed on the frictional strength of soil for ground reinforcement (piles, nailing),
few have been carried out on frictional resistance reduction. Therefore, field monitoring appears to be
essential for gaining accurate experimental data and improve prediction methods.

2. The case study of Martina Franca (Italy)

The city of Martina Franca (Ta) is located in Puglia, a region in the South East of Italy (Figure 1).
Renewal of the city sewer system is currently under construction. A new duct is to be constructed parallel
to an existing pipe which was unable to drain the flow during heavy rainfalls. The new line will drain the
waste water while rainfall will be conveyed to the existing pipe.
The project involves a section of approximately 200 m of length, to be excavated by making effective use
of trenchless technology. Excavation had to take place under a main road where heavy traffic conditions
are present all day. At the same time, the duct is to be installed at a depth between 5 to 9 m. In order to
limit interference with the surface activities, the decision was to jack a DN 600 mm gres pipe by using a
Lovat mts1000 microtunneller. One jacking shaft was excavated and two reception shafts were designed
to be constructed at the far ends of the line so that the tunnel could be excavated in two sections, using
twice the launching shaft (Figure 2).
The site is located in the central area of the city of Martina Franca which lies on a hill of limestone, locally
known as Calcare di Altamura (Altamura limestone). It consists of fairly homogeneous limestone with lo-
cal inclusions of micrite. An horizontal bedding plane is clearly visible and the rock mass is heavily
jointed. No water is present at the depth where the excavation takes place. Only rainfall water may occa-
sionally flow along the joints.
During construction of the first section, at chainage 50.68 m, the machine experienced very low advanc-
ing rate and the jacking forces reached 1.8-2.0 MN, making it unable to proceed. The microtunneller had
to be retrieved by the excavation of a new shaft and the section was completed by cut and cover. Con-
cern on the diggability of the rock mass started to take place at this time and some laboratory tests were
commissioned to various laboratories.

Figure 1 – The city of Martina Franca is located in the South East of Italy

Shaft 2 Section 2

Section 1
section by cut and cover
section by microtunnelling

Jacking shaft
Reception shaft Reception shaft

Section 1 Section 2

392 m a.s.l. filling material

heavily fractured limestone
fractured limestone
limestone with micrite inclusions

Figure 2 – Plan view and geological section of the sewer system

The same machine was set up again to excavate the second section, in the S direction, from the same
jacking shaft. After only 3 m of excavation, the cutting head got stuck and was unable to proceed. Obser-
vation of the rock mass conditions revealed that at the depth where excavation was taking place, lenses
of micrite limestone were dispersed in the rock mass, increasing its strength considerably. Other labora-
tory tests were then commissioned, in order to improve the information on the strength characteristics of
the rock mass. The machine was removed and an Herrenckhnet AVN800A microtunneller was employed
to complete the job. This second microtunneller would excavate a slightly larger tunnel (1120 mm) and
was equipped with disks on the cutting head. High jacking forces were registered also for this second mi-
crotunneller as will be shown in the following but given the new cutting tools the rock mass could be bro-
ken up and the section completed.

3. Geotechnical characterisation of limestone

As described above the geotechnical site characterisation was performed in different steps.
• At the design stage, information on the geotechnical characteristics of the ground was based on
borehole drilling and geophysical investigations.
• After the first stoppage of the Lovat microtunneller, a set of unconfined compression tests on cy-
lindrical and cubic samples were carried out.
• When the machine got trapped the second time, further investigations took place. Geomechanical
mapping, with the aim to characterise the rock mass discontinuities, together with a new set of
laboratory tests was performed.

3.1 Geological and geomechanical mapping

The rock mass is characterised by two vertical joint sets and one bedding plane, almost horizontal. A
th
structural mapping of the rock mass took place on November 17 2003 at two different locations, as
shown in Figure 3:
- an outcrop along the road near to the site,
- the ESE wall of the jacking shaft.
For both sites horizontal and vertical traverses were mapped in detail.

Figure 3 – Sites where mapping took place: outcrop and jacking shaft’s ESE wall

Table 1 – Characterisation of the joint sets*

Opening Spacing Persistence JRC
DIPDIR DIP x σ x σ x σ x σ Water Filling
Joint set [°] [°] [mm] [cm] [cm] [cm] [m] [m] [-] [-]
K1 243 80 0.1-1 0.5 14 12 1,5 0,43 14 2 Absent Absent
K2 326 78 0.1-1 0.5 17 11 1,2 0,51 17 1 Absent Absent
K3 130 11 0.1-1 0.5 35 29 2,6 0,88 14 2 Absent Absent
* JRC = Joint Roughness Coefficient, x = mean value, σ = standard deviation.
Data from mapping allowed to identify three different joint sets, as given in Table 1. Discontinuities are
mainly closed, slightly altered, filling and water are absent. Evidence of percolation is sometimes visible.
Dispersed in a relatively homogeneous limestone, lenses of micrite inclusions, from 5 to 15 cm in thick-
ness, are locally observed and called for a deeper insight into rock mineralogy.

3.2 Intact rock properties

Several laboratory tests were carried out on samples taken at different stages. A first series of uncon-
fined compression tests were performed between November 2002 and March 2003 and involved sam-
ples taken from boreholes as well as from the shaft. Further testing on samples from the jacking shaft
was performed at the DIPLAB of the Politecnico di Torino in December 2003. This series of laboratory
tests included the determination of the following:
• natural unit weight,
• mineralogical constituents,
• Knoop hardness,
• point load strength,
• unconfined compression strength,
• triaxial compression strength.
Laboratory tests were performed both on samples of limestone as well as of micrite inclusions. The
physical properties are listed in Table 2. By X-ray diffraction, no differences in mineralogy between the
inclusions and the surrounding limestone were evidenced.
The higher strength of the micrite inclusions was confirmed by point load tests, as well as unconfined
compression tests, also given in Table 2. This different behaviour relates to differences in structure, hav-
ing limestone a 40% higher porosity compared to fine micrite inclusion.
Hardness was determined by means of Knoop tests. The HK (Hardness Knoop) value was again 40%
higher in the case of the inclusions and the hardness uniformity index was 1.05 for inclusion and 1.37 for
limestone revealing a higher homogeneity for micrite and higher spatial variability in hardness for lime-
stone.
No triaxial tests were performed on the inclusions, because the samples size did not allow preparing ap-
propriate specimens. Therefore strength parameters for intact rock were determined only for the homo-
geneous limestone by means of triaxial tests on cylindrical specimens according to the ISRM recom-
mendations (ISRM 1983). An example of a triaxial test result is given in Figure 4, where strain is given by
means of both local and external measurements. Data points and failure envelope in terms of the Hoek &
Brown (1980) failure criterion are shown in Figure 5.

Table 2 - Results from laboratory tests

Micrite
Limestone
inclusion
Calcium carbonate content [%] 98.5 98.9
3
Natural unit weigth γn [kN/m ] 21.5 22.7
Porosity n [-] 20.6 14.6
Point Load Strength Index IS [MPa] 2.1 2.3
Unconfined compressive strength σci [MPa] 30.6 43.3
Hardness Knoop HK [MPa] 1300 1800
Tensile strength σt [MPa] 7.7 -
mi parameter [-] 3.7 -
Local secant Young modulus at 50% load Es,50% [GPa] 24.8 -
Local tangent Young modulus at 50% load Et,50% [GPa] 21.8 -
External secant Young modulus at 50% load Es,50% [GPa] 7.5 -
External tangent Young modulus at 50% load Et,50% [GPa] 11.2 -
Tangent Poisson’s ratio ν [-] 0.31 -
Secant Poisson’s ratio ν [-] 0.27 -
 20
Local
volumetric

16
Local
Deviator stress [MPa]

Local axial
radial
12
External
axial

0
-0,2 -0,1 0,0 0,1 0,2 0,3 0,4 0,5
Strain [%]
Figure 4 – Example of a triaxial test on limestone and specimen equipped with local transducers

100

σ 1 = σ 3 + mi ⋅ σ ci ⋅ σ 3 + s ⋅ σ ci2
mi = 3.72 σci = 30.40 s=1

σ1 σ3
0,0 36,00
0,0 22,56
σ 1 [MPa]

0,0 29,99
0,0 25,62
50 0,0 24,81
0,0 34,81
0,0 30,01
0,0 36,10
0,0 36,58
0,0 29,71
5,0 50,24
3,0 21,31

0
-10 0 10 20 30 40
σ3 [MPa]

Figure 5 – Hoek & Brown failure criterion for limestone

3.3 Rock mass properties

The rock mass has been classified by both the RMR (Bieniawski, 1989) and GSI index (Hoek 1994,
Hoek et al. 1995). To estimate the RMR index, the orientation factor was not taken into account and the
coefficient for water presence was set to zero so that strength and deformability parameters pertain to an
“intrinsic” condition. Because of the variability in the assessment of the indices, the RMR value is shown
to vary between 52 and 72. This RMR index, indicated as RMR'89, can be correlated to GSI using well
known correlations:

GSI=RMR89 - 5 valid for RMR89 > 25/

Thus, the GSI values are in the range 47-67.

The deformation modulus (Ed) for the rock mass was determined by using the following empirical correla-
tion with GSI (Hoek & Brown 1997):

GSI −10
σci
Ed = ⋅ 10 40
for σci < 100 MPa/
100

giving a value of 8.8 GPa.

Hoek & Brown strength parameters mb, sb for the rock mass (given in Figure 6) were determined versus
the GSI index. Strength parameters for the Mohr-Coulomb criterion were also determined.

35

(
σ1 = σ3 + mb ⋅ σ ci ⋅ σ 3 + s b ⋅ σ 2ci )
α

30
mb = 0.83 α = 0,5 sb = 0,0094

25
σ1 [MPa]

20

15

10 Mohr-Coulomb failure envelope

Hoek & Brown failure envelope
5

0
-5 0 5 10 15 20
σ3 [MPa]
Figure 6 – Hoek & Brown and Mohr-Coulomb failure criterions for the limestone

4. Excavation monitoring
The excavation process in both sections 1 and 2 was monitored continuously during operation. As de-
scribed previously, two different microtunnellers were adopted: a Lovat mts1000 and an Herrencknecht
AVN800A. The first machine was equipped with scrapers and conical picks while the second one had
disk cutters mounted on the cutting head (Figure 7).
The jacking force, with respect to the jacking length, measured for the Lovat mts1000 during excavation
of the first section of the line, is given in Figure 8. Monitoring data were recorded starting from 5 m of
drive. As shown in the diagram, the dotted (red) line is characterised by two slopes. During the first 15 m
the jacking force remains constant. This denotes that the jacking force, at this distance, may be mainly
attributed to face resistance. Following this, the diagram shows a steep rise, with many peaks indicating
an increase of the friction forces. Therefore two mean lines were obtained. The jacking forces reached a
limit value of 2 MN for the jacks at around 50 m drive, causing the operators to stop jacking.
The diagram of rotation per minute (RPM) of the cutting head is shown in Figure 9 versus the length of
drive superimposed on the diagram of pressure in the head drive. Both the parameters are seen to be
relatively constant along the drive and show that the cutting tools were able to correctly break up rock
along the whole length. The high increase of the gradient of the jacking forces after 15 m has to be re-
lated to other reason than cuttability.
After stoppage and rescuing, the Lovat microtunneller was set up in the jacking shaft again, to start ex-
cavation of section 2. In this case, only 3 m of advancement were possible before rotation of the head
stopped. As previously mentioned above, inclusions of micrite limestone were encountered at this time.
The tools of the cutting head were not able to disintegrate the rock, especially the micrite, and the head
could not rotate appropriately. The decision, in order to complete the job, was to change the microtun-
neller tools. To this extent a new microtunneller (Herrencknecht AVN800A) was placed in the jacking
shaft. Its cutting head was equipped with six multiple carbide-button ring disk cutters. Thanks to the new
cutting tools the rock mass could be broken up and excavation proceed again. However, difficulties were
encountered also with this second microtunneller, especially with the removal of mucking, which was fine
and sticky and caused several delays.

Figure 7 – Cutting head and positioning in the jacking shaft for the Lovat mts1000 and cutting head for
the Herrencknecht AVN800A.

The jacking forces monitored for section 2 are shown in Figure 10. For the first few meters these are very
low since the microtunneller was jacked in the 3 m long tunnel previously excavated and the work con-
sisted in slightly enlarging the tunnel section. Again two slopes can be recognised. The first one, up to
70 m advance, is mainly caused by face resistance. The second one, much steeper due to the develop-
ment of frictional forces, can be recognised, however with a local drop between 100 to 110 m advance.
Some final considerations can be drawn from the observation of the monitoring data:
• The face resistance is higher for section 2, compared to section 1. This is a clear indication of the
higher work required by the microtunneller in presence of the micrite inclusions which have higher
hardness and strength.
• Both sections are characterised by two rate of increase of the jacking force: a constant force for
the first meters followed by a steep slope. The change in slope appears after 15 m for section 1
and after 70 m for section 2. This behaviour shows that both pipes were jacked in a fairly stable
tunnel for the first part of the launch. Then frictional resistance increases probably because of in-
stabilities at the tunnel contours and fractured rocks load the pipe increasing the overall jacked
weight.
 2,5

2,0
y = 0,039x - 0,2258
Jacking force [MN]

1,5
y = 0,372

1,0

0,5

0,0
0 5 10 15 20 25 30 35 40 45 50 55
Length of drive [m]
Figure 8 – Measured jacking force versus length of drive for the Lovat mts1000

30 350

300
25

Pressure head drive [bar]

250
20
RPM [min-1]

200
15
150

10
100

5
50

0 0
0 5 10 15 20 25 30 35 40 45 50 55
Length of drive [m]
Figure 9 – Measured rotation per minute (RPM) and pressure at head drive versus length of drive
for the Lovat mts1000
 3,00

y = 0,0217x - 0,7395
2,50
Jacking force [MN]

2,00
y = 0,0025x + 0,5664

1,50

1,00

0,50

0,00
0 10 20 30 40 50 60 70 80 90 100 110 120
Length of drive [m]
Figure 10 – Measured jacking force versus length of drive for the Herrencknecht AVN 800

5. Prediction of jacking forces

Prediction of jacking forces implies the determination of the face resistance and friction forces along the
pipe. To this end, two different hypotheses can be made (O’Reilly & Rogers 1987). The first hypothesis
considers that the ground is in full contact with the outside area of the pipe, and is therefore loaded by
the surrounding ground. Determination of the load on the pipe can be done by using Terzaghi’s model or
by more accurate numerical analyses. When dealing with a rock mass, the issue is whether a continuum
or a discontinuum model of the ground is to be applied (Barla & Barla 2000). The second hypothesis as-
sumes that the excavated tunnel is stable and the average frictional resistance is only related to the
weight of the pipe.
With reference to the present case study, if one computes the displacements occurring around the tun-
nel, according to the convergence-confinement method (Panet, 1995), based on the rock mass proper-
ties given previously in this paper, the rock mass remains in the elastic state. The maximum displace-
ments computed are between 10 and 15 mm, depending from the size of the excavation. In both cases
the overcut between the ground and the pipe does not close. The pipe lies on the tunnel invert as shown
in Figure 11, for both tunnel geometries.
20
Ø
11

11
5
Ø8

10
2

Ø
Ø8
0
0

Ø600 Ø760 Ø800 Ø1040

Figure 11 – Tunnel cross section for sections 1 and 2

Based on these results, the hypothesis of stable tunnel, with friction forces only due to pipe’s weight,
seems to be more appropriate and the total jacking force to allow for pipe installation (Ptotal) may be com-
puted as:

Ptotal = Phead + Pfriction

where Phead is the force at the face of the microtunneler and Pfriction is the frictional resistance of the pipe
train along the tunnel.
Estimation of the resistance at the head is a difficult task. A detailed study of the tools-rock interaction
problem of the two microtunnellers, in line with procedures appropriate for tunnel boring machines,
brought the authors to estimate the face resistance to be of the order of 350 kN for the Lovat mts1000
and of 550 kN for the Herrencknecht AVN800A (Hurt & Laidlaw 1979, Okubo et al. 2003). This is shown
to be in good agreement with the monitored values.

3
Prediction (δ = 1/2 φ)
2,5 Prediction (δ = 2/3 φ)
Prediction (δ= φ)
Jacking force [MN]

Field data
2

1,5

0,5

0
0 5 10 15 20 25 30 35 40 45 50 55
Length of drive [m]

3
Prediction (δ = 1/2 φ)
2,5 Prediction (δ = 2/3 φ)
Prediction (δ = φ)
Jacking force [MN]

Field data
2

1,5

0,5

0
0 10 20 30 40 50 60 70 80 90 100 110 120
Length of drive [m]
Figure 12 – Computed and measured jacking forces for the two sections
The friction forces (Pfriction) can be computed as follows:

Pfriction = µ ⋅ W

where:
µ = friction coefficient;
W = weight of pipe.
The friction coefficient µ is a function of the angle δ of pipe-soil friction. Values of δ should be between ϕ
and 1/2 ϕ or 2/3 ϕ, for most geomaterials. Based on the above, the computed jacking forces for the two
sections are compared in Figure 12 to the monitored data. A reasonable agreement between predicted
and computed values is achieved for the first part of the drive, in both cases. For the remaining part of
the drive, both on section 1 and section 2, the rate of increase of friction forces is much higher than pre-
dicted. As mentioned earlier this behaviour is shown not to be related to the boreability of the rock mass,
as the face resistance is well predicted, but may be caused by instabilities occurring at the tunnel con-
tour. If this is the case the first hypothesis may not be sufficient and computation of jacking forces should
account for the weight of unstable rock blocks. To clarify this point a further insight into the rock mass re-
sponse to excavation is necessary.

6. Continuum versus discontinuum modelling and prediction of jacking forces

In the above computations, the stress ratio (Ko) of the rock mass has been considered equal to one, i.e.
vertical effective stress equal to horizontal effective stress. The stress ratio also applies to total stresses
being the rock mass dry. If consideration is given to the geological history of the site it can be foreseen
that the stress ratio may likely be lower than one. No field measurement is available but considerations
based on previous experience may allow expecting values between 0.1 and 0.4.
Low horizontal stresses in a fractured rock mass as that under study may allow one to anticipate opening
of the vertical joints. If we remain in a continuum framework, this phenomenon may be considered artifi-
cially by reducing the tensile strength of the rock mass and applying the reduced stress ratio in a finite
difference numerical model. To get a deeper insigth, parametric numerical analyses of a circular tunnel
were run by using the Flac code (Itasca 2003). An appropriate mesh was built, according to the geometry
of the problem and analyses were run to simulate the excavation of the tunnel varying two parameters:
the stress ratio and the tensile strength of intact rock. The results of these analyses are shown in Figure
13 where the effect of both parameters can be seen. By decreasing the Ko ratio, elements around the
tunnel fail. These failure zones are localised at the crown of the tunnel and are shown to increase as the
tensile strength and the stress ratio decrease.
1,4
Heigth of failure zone at crown [m]

failure zones Ko = 0.1

1,2 at the crown Ko = 0.2
Ko = 0.3
1

0,8

0,6

0,4

0,2

0
0 0,02 0,04 0,06
Tensile strength [MPa]
Figure 13 – Continuum modelling results
The results obtained are a clear indication that loose zones around the tunnel are likely to develop and
blocks may slide onto the pipe, increasing the force necessary to jack it into the ground.
A better description of ground behaviour is achieved by accounting for the discontinuum nature of the
rock mass, by discontinuum modelling. To this end, the distinct element method and the Udec code
(Itasca, 2003) were used. A distinct element model was built according to the geometry of the site, with
appropriate boundary conditions. Gravity loading was applied to the model with a stress ratio Ko equal to
0.3. Boundary conditions are shown in Figure 14.
The rock mass is given an average GSI index equal to 58. The model was built by considering the aver-
age joints spacing. Intact rock deformability and strength parameters were assigned to rock blocks and
joints based on the geotechnical characterisation above. The joints, which were assumed to follow the
Mohr-Coulomb criterion, are given the properties listed in Table 3.

Table 3 – Parameters for discontinuities in the analyses with GSI = 58*

kn ks φ
Joint set [GPa/m] [GPa/m] [°]
K1 102.8 39.1 33.0
K2 87.4 33.2 36.0
K3 41.3 15.7 29.7
* kn = normal stiffness, ks = shear stiffness, φ = friction angle.

Following stress initialization, the analysis was run to simulate complete excavation of the tunnel by ap-
propriate removal of rock blocks. Equilibrium is reached when no movement is recorded in the model.
Figure 15 shows the situation after excavation. Green vectors indicate displacements while red lines
show the joint opening. Colour scale, from green (minimum) to red (maximum) indicate vertical velocity
value.

Ko = 0.3

Figure 14 – Udec model and boundary conditions for GSI = 58

 Figure 15 – Block movements and velocity vectors after excavation (GSI = 58)

Only a few blocks are shown to be mobilised. As a consequence, the friction forces computed on the ba-
sis of the rock load are still not able to justify the field performance.
The results above are obtained for a rock mass with GSI equal to 58. It has to be noted that, for the case
study, conditions of the rock mass may vary along the drive. If the rock mass quality decreases, further
blocks may fall, due to stress release.
Direct observation during cut and cover operations, to rescue the Lovat mts1000 microtunneller after
stoppage, revealed a heavily fractured rock mass at that chainage. In some way this was foreseen by the
seismic investigation and borehole interpretation performed to draw the geological section of Figure 2.
Based on these considerations, the decrease of the rock mass conditions was simulated by running fur-
ther discontinuum analyses. The GSI index was scaled down to 38 by appropriately varying the input pa-
rameters. The new input data for the discontinuities are given in Table 4.

Table 4 – Parameters for discontinuities in the analyses with GSI = 38*

kn ks φ
Joint set [GPa/m] [GPa/m] [°]
K1 32.6 12.4 27.1
K2 27.7 10.5 29.2
K3 13.1 4.9 24.9
* kn = normal stiffness, ks = shear stiffness, φ = friction angle.

Results of the second analysis are shown in Figure 16 where green arrows indicate velocity vectors and
red lines show joint opening. Again the colour plot shows vertical velocities.
If a comparison is made with the results of the previous analyses, greater displacements take place
along the joints. In particular, an area of unstable blocks is shown to develop at the tunnel crown and
sidewalls. The area of the mobilised blocks is shown to be in reasonable agreement with the result of the
continuum analyses depicted in Figure 13.
It is evident that if the rock mass is more fractured, the forces needed to jack the pipe will increase pro-
portionally, due to the effect of failure at the tunnel surround. These conditions are likely to have occurred
for the case study described, where both along section 1 and 2 the rock mass quality may have become
worse. Moreover, this was observed directly when the first microtunneller was retrieved by cut and cover.
Block instability gives rise to an increase of friction forces along the pipe. At the same time sidewalls in-
stability may lead to misalignments of the pipe train during jacking. If these effects are considered in the
computation of jacking forces, then a better comparison with field data may be obtained as shown in Fig-
ure 17. Prediction is in good agreement with monitoring for section 2. It appears that, for section 1, the
higher forces monitored cannot be referred only to unstable rock blocks. One may argue that additional
thrust was applied by the jacks in order to improve the advancement.
 Figure 16 – Block movements and velocity vectors after excavation (GSI = 38)

Prediction (δ = φ)
2,5 Field data
Jacking force [MN]

1,5

0,5

0
0 5 10 15 20 25 30 35 40 45 50 55
Length of drive [m]

3
Prediction (δ = 2/3 φ)
Prediction (δ = φ)
2,5 Field data
Jacking force [MN]

1,5

0,5

0
0 10 20 30 40 50 60 70 80 90 100 110 120
Length of drive [m]

Figure 17 – Computed and monitored jacking forces for section 1 and 2

7. Conclusions
An Italian case study where a microtunneller got stuck during operation, due to local instability in the rock
mass around the tunnel, has been described. It has been shown that design based on conventional con-
tinuum models did not allow predicting completely the jacking forces measured in the field. The excava-
tion was therefore back analysed by means of numerical analyses with the distinct element method and
the UDEC code. Only by taking into account the discontinuum behaviour of the rock mass it was possible
to predict the tunnel response to excavation, which led to the high jacking forces recorded and were re-
sponsible for the first stoppage.
The procedure adopted in the paper can be followed if a complete geotechnical investigation of the rock
mass is available at the design analysis stage. The case study described is a clear example of how cor-
rect and complete information of the geotechnical aspects of the site is essential at the design stage and
may allow civil engineers to gain useful prediction of complex problems such as microtunnelling opera-
tions.

8. References
Barla, G., Barla M., 2000, Continuum and discontinuum modelling in tunnel engineering, Gallerie e
Grandi Opere Sotterranee, 61, Pàtron Editore, Bologna.
Bieniawski, Z.T., 1989, Engineering Rock Mass Classification, 251, Wiley, New York (USA).
Hoek, E., 1994, Strength of rock and rock masses. ISRM New Journal, 2(2), 4-16.
Hoek, E., Kaiser, P.K., Bawden, W.F., 1995, Support of Underground Excavations in Hard Rock, 215,
Balkema, Rotterdam (The Netherlands).
Okubo, S., Fukui, K., Chen, W., 2003. Export system for applicability of tunnel boring machines in Japan,
Rock Mechanics and Rock Engineering, 36(4), 305-322.
Hurt, K.G., Laidlaw D.D., 1979, Laboratory comparison of three rock cutting tools, Tunnels & Tunnelling,
May 1979.
Hoek, E., Brown, E.T., 1980, Underground excavations in Rock, 527, Inst. Min. Metall., London (UK).
Hoek, E., Brown, E.T., 1997, Practical estimates of rock mass strength. Int. J. Rock Mech. Min. Sci., 34,
1156-1186.
Isrm, 1983, Suggested methods for determining the strength of rock material in triaxial compression: re-
vised edition, Int. J. Rock Mech. Min. Sci. & Geomech. Abstracts, 20, n. 6, 283-290.
Itasca, 2001, Flac (Fast Lagrangian Analysis of Continua) ver 3.4, Itasca consulting group, Minneapolis,
U.S.A.
Itasca, 2003, Udec (Universal Distinct Element Code) ver 3.1, Itasca consulting group, Minneapolis,
U.S.A.
O’Reilly M.P., Rogers C.D.F., 1987, Pipejacking forces, Proceedings of International Conference on
Foundation and Tunnels, Edinburgh (UK).
Panet, M., 1995, Calcul des tunnels par la méthod convergence-confinement, Presses de l’éecole natio-
nal des Ponts e Chassées, Paris.

9. Aknowledgements
The Authors would like to acknowledge the help of Mr. Marco Capelli. The Authors also would like to ac-
knowledge the permission of Vepo S.r.l. and Pfeiffer Italia S.p.A. to publish the monitoring data and pho-
tographs of the site.
The work described in this paper was carried out with the financial support of the Italian Ministry for Uni-
versity and Research (MIUR) as part of the Research Programme “Mechanised excavation of tunnels”,
co-ordinated by Prof. G. Barla, Politecnico di Torino.