The NTNU Prediction Model: A Tool for Planning and Risk Management in Hard Rock TBM Tunnelling

Abstract

TunIn the tunnelling industry, the prediction of TBM performance and costs is an important issue for planning and risk management at the design stages. The NTNU prediction model is an objective tool that has been used by several parties in the tunnelling industry. The model can be used for estimating the excavation rate and costs, as well as planning and risk assessment of hard rock TBMs projects. The model is mainly used in the planning stage for hard rock TBM tunnelling projects. However during the last few years, the model has been used as a tool for the resolution of disputes between clients and contractors during the construction stage. In this paper, a study of the NTNU prediction model for hard rock TBMs is carried out in a case of 3.4 m diameter open hard rock TBM. The NTNU prediction model is validated for the present project as a good tool for project planning, risk management and useful tool in assessing claims.

Full text

Proceedings of the World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil.
1
1 INTRODUCTION
Underground development using the Tunnel
Boring Machine (TBM) method is increasing
widely with a large number of underground
projects excavated in hard rock. TBM
performance is a key factor when it comes to
have the highest influence on the completion of
TBM projects on time and cost. In TBM
performance prediction, the penetration rate is
the main factor for estimating advance rate,
cutter life and excavation costs in hard rock
tunnelling. Accurate predictions avoid time
delays and budget overruns resulting in a
control of the risk for tunnelling projects.
The penetration rate of hard rock TBMs is an
interaction of the rock mass properties and the
machine parameters.
Several prediction models and updates of the
existing models have been developed in the last
few decades. Some of them are Graham (1976),
Farmer and Glossop (1980), Büchi (1984), CSM
model (Rostami and Ozdemir, 1993, Rostami,
1997), Luleå University of Technology (Nelson
et al., 1994), NTNU model (Bruland, 1998),
QTBM (Barton, 2000), RME (Bieniawski, 2006)
Gong and Zhao, 2009) or Hassanpour et al.
(2009).
Presently, the Colorado School of Mines
(CSM) model (1993, 1997) and the Norwegian
University of Science and Technology (NTNU)
model (1998) are the prediction models most
widely used for hard rock TBM tunnelling
application.
The NTNU prediction model for hard rock
TBMs combines rock boreability properties and
TBM parameters in order to determine the main
factors influencing the penetration rate
predictions. Predictions of advance rate, cutter
wear and excavation costs are also achieved
using the NTNU prediction model.
The purpose of the paper is to analyse the
NTNU prediction model as a tool for planning,
and risk management.
In order to carry out the goal, an analysis of
the penetration rate prediction accuracy by
using the NTNU prediction model in a recently
finished project case of a 3.4 m diameter open
hard rock TBM was performed. Estimation of
advance rate, cutter wear and costs are not
reported in the present paper.
The NTNU Prediction Model: A Tool for Planning and Risk
Management in Hard Rock TBM Tunnelling.
F. J. Macias, P. D. Jakobsen and A. Bruland
Norwegian University of Science and Technology, Trondheim, Norway.
S. Log
The Robbins Company, Oslo, Norway
E. Grøv
SINTEF Building and Infrastructure, Rock Engineering, Trondheim, Norway
ABSTRACT: In the tunnelling industry, the prediction of TBM performance and costs is an
important issue for planning and risk management at the design stages. The NTNU prediction model
is an objective tool that has been used by several parties in the tunnelling industry. The model can be
used for estimating the excavation rate and costs, as well as planning and risk assessment of hard
rock TBMs projects. The model is mainly used in the planning stage for hard rock TBM tunnelling
projects. However during the last few years, the model has been used as a tool for the resolution of
disputes between clients and contractors during the construction stage. In this paper, a study of the
NTNU prediction model for hard rock TBMs is carried out in a case of 3.4 m diameter open hard
rock TBM. The NTNU prediction model is validated for the present project as a good tool for project
planning, risk management and useful tool in assessing claims.

Proceedings of the World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil.
2
Predictions of TBM penetration rate give
reasonable standard errors as well as good
reliability and validity for the case studied.
A research project regarding the NTNU
prediction model is underway to be reported
later. Preliminary data from the discussion of
the RPM influence on then TBM penetration
rates are shown.
2 THE NTNU PREDICTION MODEL FOR
HARD ROCK TBMS
2.1 Introduction
The philosophy of the NTNU prediction model
is to combine the decisive rock properties and
the relevant machine parameters.
Several steps are involved in the NTNU
prediction model for hard rock TBMs in order
to estimate time and cost for tunnel excavation.
‐ Net penetration rate
‐ Cutter life
‐ Advance rate
‐ Excavation costs
The model has had a successive development
since the first version in 1976 by the NTNU
(former NTH).
Table 1 shows the successive editions of the
NTNU prediction model to date.
Table 1: History of the NTNU prediction model for hard
rock TBMs
EDITION YEAR
1st edition 1976
2nd edition 1979
(published in 1981)
3rd edition 1983
4th edition 1988
5th edition 1994
6th edition 1998
The last prediction model edition (Bruland,
1998) is based on data from almost 250 km. of
tunnels
2.2 Parameters of the model
Net penetration rate and cutter life depends on
rock properties and machine parameters.
The rock parameters consists of intact rock
and rock mass parameters. The rock parameters
are combined in a rock mass boreability
parameter, the equivalent fracturing factor (kekv)
while the machine parameters are combined into
one machine parameter, the equivalent thrust
(Mekv).
Table 2 shows the rock properties and
machine parameters influencing the net
penetration rate.
Table 2: Rock and machine parameters influencing the net
penetration rate
ROCK PARAMETERS MACHINE
PARAMETERS
Intact Rock
Parameters
Rock Mass
Parameters
Drilling Rate
Index, DRITM Rock Mass
Fracturing
Factor (ks)
TBM diameter
Cutter diameter
Porosity
Number of cutters
Average cutter
thrust
Average cutter
spacing
Cutterhead RPM
Installed cutterhead
power
Since the paper is not related to cutter wear,
only the main concepts of cutter wear estimation
are shown below for general background.
The cutter wear model is based on time
dependent abrasion of the cutter rings. The
model is entirely based on field data and
corresponding rock samples tested in the
laboratory.
Cutter life in hours is equivalent to the cutter
life in rolled distance (km/cutter) for a given
cutterhead RPM. The cutter life in hours is
combined with the penetration rate (m/h) and
the TBM diameter to calculate the cutter life in
m/cutter and sm3/cutter (solid cubic meters
excavated by cutter).
The cutter wear depends on the following
rock properties and the machine parameters
given in Table 3.
Table 3: Rock and machine parameters influencing the
cutter wear (Bruland, 1998)
Rock Mass Parameters Machine parameters
Cutter Life Index,
CLITM
Number of cutters
Cutter diameter
Rock Quartz
content (%)
TBM diameter
Cutterhead RPM

Proceedings of the World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil.
3
2.2.1 Rock Parameters
The Drilling Rate Index (DRITM) expresses the
drillability or boreability of the intact rock.
DRITM is evaluated on the basis of two
laboratory tests, the Brittleness Value (S20) test
and the Sievers’ J-Value (SJ) miniature drill
test.
DRITM may be defined as brittleness or the
ability to be crushed by repeated impacts,
corrected for the surface hardness determined by
the SJ (Dahl et al. 2012). The DRITM is an
indirect measure of the required breaking work
and a good representation of the rock breaking
process under a cutter.
Figure 1 shows the variation found in DRITM
for some rock types.
Figure 1: Recorded Drilling Rate Index (DRITM)
for some rock types (Bruland, 1998)
The porosity of the rock shows a clear
influence on the penetration rate (Bruland,
1998). A separate correction factor for the rock
porosity is hence included in the model.
The influence of the porosity can be
explained by the pores acting as crack initiators
and amplifiers of the crack propagation.
Rock mass fracturing is found to be the rock
mass parameter of largest influence on the net
penetration rate of TBMs in hard rock
conditions (Bruland, 1998).
The rock mass fracturing factor (ks) considers
the simultaneous influence of the degree of
fracturing and the orientation of the plane of
weakness in the rock mass fracturing.
2.2.2 Machine parameters
The main machine parameter boring in hard
rock is the average cutter thrust. When the thrust
is increased, the cutter will indent deeper into
the rock surface and therefore transmit the
energy from the cutterhead to the rock more
efficiently.
Gross average cutter thrust is used in the
NTNU prediction model. This means that the
total cutterhead thrust is divided by the number
of cutters on the cutterhead and also averaged
over the time.
Applied thrust depends on the rock mass
fracturing. It is hence recommended to use
different gross average cutter thrust depending
on the degree of fracturing of the rock. Higher
gross average cutter thrust for lower Ks values
and lower gross average cutter thrust for higher
Ks values.
Other input of the model is the cutter
diameter. Increasing the cutter diameter gives
rise to increasing the applicable cutter thrust.
A correction factor for the cutter diameter is
included in the model. It is related to cutter ring
edge width since variations in the cutter
diameter means variations of the edge width of
the standard rings. The NTNU model does not
consider the cutter edge width as an independent
parameter.
Other parameters included in the model are
the average spacing of the cutters on the
cutterhead. The average cutter spacing means
the cutterhead radius divided by the number of
cutters on the cutterhead.
The model does not consider the possible
influence of the TBM cutterhead shape, e.g. flat
and domed.
Most of the model database is associated
with TBMs which have a cutterhead RPM
according to the maximum rolling velocity of
the outer gauge cutter. Therefore the model does
not consider the influence of the cutterhead
RPM on the penetration. The cutterhead RPM
has an influence on the penetration rate per
revolution and a first indication is given by
Bruland (1998).
2.3 Penetration rate model
As already mentioned, the philosophy of the
NTNU prediction model is to combine the
decisive rock parameters and the relevant
machine parameters, Bruland (1998).
The factor for the rock parameters is the
equivalent fracturing factor (Kekv), Equation 1:

Proceedings of the World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil.
4
porDRItotsekv kkkk   , (1)
where ks-tot is the total rock mass fracturing
factor, kDRI is the correction factor for DRI of
the rock and kpor is the correction factor for
porosity of the rock
The factor for the TBM parameters is the
equivalent cutter thrust factor (Mekv) in
kN/cutter, equation 2:
adBekv kkMM  , (2)
where MB is the gross average cutter thrust (The
thrust which will be used), kd is the correction
factor for the cutter diameter and ka is the
correction factor for average cutter spacing
The penetration rate estimation model is
based on normalized penetration curves
according to Bruland (1998).
Penetration curves are resultant from
penetration tests performed with TBMs during
boring. Figure 2 shows a general progress of a
penetration curve.
Figure 2: General progress of a penetration curve
(Bruland, 1998)
Equation 3 relates to the penetration curve. It
is a good fit for a wide range of penetration
tests.
b
ekv
oM
M
i








1
, (3)
where io is the basic penetration rate (mm/rev),
Mekv is the equivalent cutter thrust factor
(kN/cutter), M1 is the critical cutter thrust in
kN/cutter (It is the necessary thrust to achieve 1
mm/rev) and b is the penetration coefficient
Critical cutter thrust (M1) and penetration
coefficient (b) have relation with the equivalent
fracturing factor (kekv), Bruland (1998).
Basic net penetration rate (Io) is applicable
for systematically fractured rock mass and is
defined as metres bored per hour. Io is a
function of the basic penetration rate and the
cutterhead rpm, equation 3:







1000
60
RPMiI oo , (4)
where Io is given in m/h.
In case of presence of “marked single joints” a
correction factor for Marked Single Joints (kesp)
is used. Equation 5 shows the net penetration
rate with marked single joints.
espoesp kII


, (5)
For high net penetration rates when boring in
fractured rock, it should be checked that there is
sufficient cutterhead power installed in the
TBM.
Besides the available torque, it also necessary
to check the system’s capacity for muck
removal in order to avoid limits on the net
penetration rate.
2.4 Output parameters
The NTNU model estimates the following
parameters:
 Net penetration rate (m/h)
 Cutter life (h/cutter, sm3/cutter)
 Machine utilization (%)
 Weekly advance rate (m/week)
 Excavation Costs (NOK/m)
The model allows developing sensitivity
analysis of influence of one or more factors on
the output parameters. For example the
geological risk can be analysed keeping
constant the TBM parameters and varying the
rock parameters.
The machine utilization derives from the
estimated boring time divided by the total
available time per shift, day or week. It is based
on the estimation of the time consumption in
h/km. Machine utilization must be analysed
from the TBM side and not for the tunnel as a
whole.
The advance rate is estimated by the net
penetration rate (m/h) and the machine
utilization (%). The model is based on averaged
data over the complete tunnel length.

Proceedings of the World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil.
5
Related with excavation costs, the cost model
is directly related to the models for penetration
rate, cutter life and advance rate. Normalized
TBM costs, consumables costs and interest rates
are considered.
The cost model is based on detailed
estimations of all excavation costs. In the 1998
version of the model, the cost and possible extra
time for rock support measures are not included.
3 FIELD DATA FOR PERFORMANCE
PREDICTION
The EIDI 2 hydropower project at Faroe Islands
has been selected to study the applicability of
the NTNU model. Detailed data of the rock
mass fracturing, laboratory indices (DRI™ and
CLI™) and the TBM performance as the
mapping facilities is excellent due to the has
been gathered. The site was chosen as the TBM
is an open gripper TBM, making detailed rock
mass mapping possible.
The geology along the 8 km tunnel consists
of basalt and sill.
Drillability testing was performed in the
Engineering Geology Laboratory at
NTNU/SINTEF according to Dahl et al. (2013).
Table 4 shows the average values of the
drillability indices for sill and basalt.
Table 4: Drillability values for the rock types
Rock Type Sill Basalt
DRITM 38 64
CLITM 15 41
Quartz content 1 % -
According to Bruland (1998), sill is classified
as a rock with a “low” DRI™ that means low
boreability or resistance to boring while the
basalt is a rock classified as “high” drilling rate
that means high boreability or a rock easily
bored.
A complete field mapping according to Bruland
(1998) was performed in a total length of 1 200
m at the EIDI 2 project. Figure 3 shows the rock
mass fracturing factor (ks) values along the
mapped sections.
Figure 3: Rock mass fracturing factor (ks) along the
mapped sections
The TBM performance data was recorded
from the boring reports. Boring time, boring
length and the cutterhead thrust are collected for
different working periods.
The TBM used for the excavation is a main
beam (open gripper) Robbins High Performance
TBM (HP). The main TBM characteristics are
shown in Table 5
Table 5: Main TBM characteristics
TBM parameters Value
TBM diameter (m) 3.4
Cutter diameter (mm) 432
Maximum cutter thrust recommended
(kN/cutter)
237
Cutterhead RPM 12.8
Installed cutterhead power (kW) 671
The cutter thrust applied (kN/cutter) is the
main machine parameter influencing the TBM
penetration rate.
It is well known that a reduction of the thrust
is made when the rock mass fracturing degree is
increasing in order to avoid excessive wear or
even damage to the cutters and possible damage
to the machine.
However, for low rock mass fracturing
degree values, the necessary thrust for boring is
higher. This is due to that thrust becomes the
main factor to increase the net penetration rate
when the intact rock properties dominate the
rock mass boreability.
Figure 4 shows the maximum recommended
by the manufacturer and achieved cutter thrust
(kN/cutter).

Proceedings of the World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil.
6
Figure 4: Maximum and recorded cutter thrusts
(kN/cutter)
It should be considered that the thrust levels
shown in Figure 4 are lower than it would be
expected. This may be because the TBM used
was manufactured in the mid 1980’s although
still in use. Anyway, these ranges of values are
in accordance with several examples of gross
cutter thrust reduction shown in Farrokh et al.
(2012).
As showed in Figure 4, the cutter thrust
applied is lower than the maximum
recommended thrust according to the cutter
capacity and rock mass degree of fracturing.
In this case, the total cutter thrust average is
almost 30% lower than the recommended
maximum thrust.
For low ks values, rock mass with few joints,
the applied thrust was 10% higher than the total
average while for high ks values, fractured rock
mass, the thrust applied was 12% lower.
In the present case after analysing the data, it
was defined a ks value of 1.3 to define low and
high ks values.
Since the applied cutter thrust is the main
factor influencing the TBM penetration rate, it
is a highly important input parameter in the
performance prediction.
A sensitivity analysis is performed in order to
determine the influence of the cutter thrust on
the penetration rate predictions.
In the approximately first 270 m the mixed
face phenomena appears. Mixed face conditions
occur when the tunnel face contains rock
sections with significantly different properties
regarding to rock boreability. Sill and basalt are
the rock types involved.
During boring in mixed-face the cutter thrust
is substantially reduced in order to avoid cutter
and machine damages. Figure 4 shows the cutter
thrust average in the mixed face, which is
almost 45% lower than the maximum
recommended. Likewise, the penetration rate is
also reduced.
Steingrimsson et al. (2002) proposed a
“mixed-face correction factor, KMF” for the
NTNU prediction model for hard rock TBMs.
The prediction along the mixed- face section
will also be analysed using the “mixed-face
correction factor (KMF).
4 ANALYSIS AND DISCUSSION OF THE
PERFORMANCE PREDICTIONS
Several predictions along the studied section are
performed for different cutter thrust
assumptions, actual cutter thrust, total cutter
thrust average and two cutter thrust levels
regarding to the ks range.
The calculations were performed by using
the NTNU prediction model software
“FULLPROF”.
Figure 5 shows the actual and predicted
penetration rates along the studied section.
Figure 5: Actual and predicted penetration rates along the
studied section
An analysis of the prediction errors for every
prediction case is carried out. Figure 6 shows
the error average, standard error, maximum and
minimum.

Proceedings of the World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil.
7
Figure 6: Average, standard error and outliers for
prediction with actual thrust, total thrust average and
thrust average regarding to the ks
Maximum error values are achieved for the
predictions using the actual thrusts recorded. In
addition, the predictions have lower validity and
reliability. Low validity means large range of
errors while low reliability means high error
average.
The total average errors for the predictions
using the total average and average with respect
to the ks level are very low and almost equal.
Standard error and outliers for the prediction
with a cutter thrust regarded to the ks level are
almost half (13%) the same values achieved for
the prediction using total cutter thrust average
(27%).
Standard errors below 30% can be
considered a good level of prediction
considering the geological uncertainty.
Figure 7 shows the actual and predicted used
cutter thrust regarded with the ks level along the
studied section. In addition, parallel prediction
for the mixed face (MF) section is estimated
using the correction factor proposed by
Steingrimsson et al. (2002).
Figure 7: Actual and predicted penetration rate using
cutter thrust average for low and high ks.
Figure 7 shows that the prediction fits well
throughout the section studied. The largest
deviations correspond to the mixed face section.
Figure 8 shows a comparison of the results
from the prediction using cutter thrust
associated with the Ks level with the actual
penetration rate values.
Figure 8: Comparative results for the prediction using the
NTNU prediction model with average cutter thrust
regarding to the ks level
Figure 8, as Figure 7, show that the
prediction values have a good fit. The fitting
does not show significant variations along the
penetration rate values.
Figure 6, Figure 7 and Figure 8 indicate that
the estimating errors are not systematic. The
prediction values are scattered around the actual
values.
The predictions values have low scattering
which means a high reliability of the prediction.
Also the predictions values have a low average
standard error which means high validity of the
prediction.
It should be considered that the TBM used
for the excavation in the studied case has a
technology in accordance with the last update
NTNU model database (Bruland, 1998). The
prediction model should be also analysed for
current hard rock TBMs.
5 CONCLUSIONS
The main conclusions are listed below:
 Average cutter thrust should be used for
performance prediction by using the
NTNU prediction model for tunnel
boring.
 A different boreability role is found for
low and high ks values. So, in order to

Proceedings of the World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil.
8
improve the reliability of the predictions,
different average of cutter thrust should
be used in the predictions according to
the rock mass fracturing (ks values).
 Similar validity is found in the
predictions using a cutter thrust average
for the total length and different thrust
level regarding to the ks range.
 Comparison of the predicted and real
penetration rate values indicate low
scattering.
 Predictions using the NTNU model in
the case of study have an acceptable
accuracy and low scattering level.
 For the present studied case the NTNU
prediction model has good reliability and
good validity.
 The NTNU prediction model is indicated
as a good tool for planning, risk
management and assessment of claims in
hard rock tunnel boring.
 It is possible to say that the Mixed Face
correction factor suggested by
Steingrimsson et al. (2002) improve
slightly the predictions for the studied
case. There is still a lack of knowledge
and understanding of tunnel boring
behaviour in mixed face sections.
6 FURTHER RESEARCH
A current research project regarding to the
NTNU prediction model is underway.
New data available from current projects will
be added in order to update the current NTNU
model edition as well as increase the application
range (for example for larger diameters, larger
cutter discs…)
In addition, an analysis of the main
parameters involved in the estimations will be
carried out. The geological mapping
methodology, influence of the petrography on
DRITM and CLITM (grain size, texture, mineral
content), influence of the cutterhead RPM on
the net penetration rate or evaluating the need of
a new abrasivity test method are examples of
this analysis.
Several tests have been performed in order to
evaluate the RPM influence on the net
penetration rates for hard rock TBMs.
Figure 9 shows the influence of the
cutterhead RPM on the net penetration rates.
Figure 9: Penetration tests with variable cutterhead RPM
and constant thrust
Figure 9 shows two penetration tests
performed in two different projects. These
penetration tests were performed keeping almost
constant the applied thrust and varying the
cutterhead RPM. Figure 9 indicates that
decreases in the cutterhead RPM, keeping
constant the thrust, give rise to increases in the
net penetration rates.
High loading rates, due to high cutterhead
RPM, does not allow dislocations, deformations
and microcracks propagation. This indicates
that, although the cutter load level is higher, the
effectiveness of the cutter indentation is lower
resulting in a lower net penetration rate.
ACKNOWLEDGEMENTS
The authors would like to thank the research
project “Future Advanced Steel Technology for
Tunnelling” (FAST-Tunn). This project is
managed by SINTEF/NTNU and funded by the
Research Council of Norway, the Robbins
Company, BASF Construction Chemicals, the
Norwegian Railroad Authorities, Scana Steel
Stavanger and BMS steel.
REFERENCES
Bieniawski von Preinl, Z.T.; Celada Tamames, B.; Galera
Fernandez, J.M. and Alvarez Hernandez, M. 2006.
Rock mass excavability indicator: new way to
selecting the optimum tunnel construction method.
Tunnelling and Underground Space Technology 21
(3-4), 237
Bruland, A. 1998. Hard Rock Tunnel Boring: Vol 1 – 10,
PhD thesis. Norwegian University of Science and
Technology (NTNU), Trondheim, Norway.
Bruland, A. 1998. Prediction Model for Performance and
Costs. “Norwegian TBM Tunnelling” Publication
NO.11: 29-34, Norwegian Tunnelling Society

Proceedings of the World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil.
9
Dahl, F.; Bruland, A.; Jakobsen, P.D.; Nilsen, B. and
Grøv, E. 2012. Classifications of properties
influencing the drillability of rocks, based on the
NTNU/SINTEF test method. Tunnelling and
Underground Space Technology 28: 150-158
Delisio, A.; Zhao, J. and Einstein, H.H. 2013. Analysis
and prediction of TBM performance in blocky rock
conditions at the Lötschberg Base Tunnel. Tunnelling
and Underground Space Technology 33: 131-142
Farrokh, E.; Rostami, J. and Laughton, C. 2012. Study of
various models for estimation of penetration rate of
hard rock TBMs. Tunnelling and Underground Space
Technology 30: 110-123
Hassanpour, J.; Rostami, J.; Khamehchiyan, M.; Bruland,
A. and Tavakoli, H.R. 2010. TBM performance
analysis in pyroclastic rocks: A case history of Karaj
water conveyance tunnel. Rock Mech. Rock Eng.
43:427-445
NTNU 2009. Fullprof software 2009. Norwegian
University of Science and Technology (NTNU),
Trondheim, Norway
Palm, A. 2011. Driftsoppfølging av TBM-prosjektet Eidi
2 Sudur. Master Thesis. Norwegian University of
Science and Technology (NTNU), Trondheim,
Norway.
Rostami, J. and Ozdemir, L. 1993. A new model for
performance prediction of hard rock TBMs.
Proceedings of the Rapid Excavation and Tunnelling
Conference 1993, Boston, pp 793-809.
Rostami, J.; Ozdemir, L. and Nilsen, B. 1996.
Comparison between CSM and NTH hard rock TBM
performance prediction models. ISDT, Las Vegas NV.
1996.
Sapigni, M.; Berti, M.; Bethaz, E.; Busillo, A. and
Cardone, G. 2002. TBM performance estimation using
rock mass classifications. Int. J. Rock. Mech. Min. Sci.
& Geomech. Abstr. Vol 39 pp 771-788
Steingrimsson, J.H.; Grøv, E. and Nilsen, B. 2002. The
significance of mixed-face conditions for TBM
performance. World Tunnelling, 9: 435-441
Yagiz, S.; Rostami, J.; Kim, T.; Ozdemir, L. and
Merguerian, C. 2010. Factors influencing performance
of hard rock tunnel boring machines. Rock
Engineering in Difficult Ground Conditions- Soft
Rocks and Karst- Vrkljan (ed). Taylor & Francis
Group, London, ISBN 978-0-415-80481-3 pp 691-700
Zare, S. and Bruland, A. 2012. Applications of
NTNU/SINTEF Drillability Indices in Hard Rock
Tunneling. Rock. Mech. Rock Eng. (2013) 46:179-187
Zhao, Z.; Gong, QM.; Zhang, T. and Zhao, J. 2007.
Prediction model of tunnel boring machine
performance by ensemble neural networks. Geomech.
Geoeng. Int. J. 2 (2): 123 - 128