Professional Documents
Culture Documents
A new hard rock TBM performance prediction model for project planning
J. Hassanpour a,⇑, J. Rostami b, J. Zhao c
a
SCE Company, Tehran, IRAN, P.O. Box 16765-3465, Iran
b
Department of Energy and Mineral Engineering, Pennsylvania State University, University Park, PA, USA
c
Ecole Polytechnique Federale de Lausanne (EPFL), Lausanne, Switzerland
a r t i c l e i n f o a b s t r a c t
Article history: Among the models used for performance prediction of hard rock tunnel boring machines two stand out
Received 12 October 2010 and are often used in the industry. They include the semi theoretical model by Colorado School of Mines
Received in revised form 27 March 2011 and the empirical model by Norwegian University of Science and Technology in Trondheim (NTNU).
Accepted 9 April 2011
While each have their strong points and area of applications, more accurate prediction has been sought
Available online 4 May 2011
by modifying one of the existing models or introduction of a new model. To achieve this, a database of
actual machine performance from different hard rock TBM tunneling projects has been compiled and
Keywords:
analyzed to develop a new TBM performance prediction model. To analyze the available data and offer
TBM performance
Rock mass boreability
new equations using statistical methods, relationships between different geological and TBM operational
Field Penetration Index parameters were investigated. Results of analyzes show that there are strong relationships between geo-
logical parameters (like UCS, joint spacing and RQD) and TBM performance parameters specially Field
Penetration Index (FPI). In this study, a boreability classification system and a new empirical chart, for
preliminary estimation of rock mass boreability and TBM performance is suggested.
Ó 2011 Elsevier Ltd. All rights reserved.
0886-7798/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tust.2011.04.004
596 J. Hassanpour et al. / Tunnelling and Underground Space Technology 26 (2011) 595–603
machine. In a similar research work, Gong and Zhao (2009) by per- geological zones of Iran and Manapouri second tailrace tunnel
forming a nonlinear regression analysis on data obtained from two were selected for this study. The main characteristics of these
tunnels excavated in granitic rock masses in Singapore developed TBM tunneling projects are summarized in Tables 1 and 2.
an empirical equation to estimate boreability of rock mass. They As can be seen in the Table 2, these tunnels have been con-
proposed a relationship between four rock mass parameters structed in different rock types including sedimentary, igneous
(UCS, brittleness, joint count number, and orientation of joints) and metamorphic rocks with a wide range of rock strength.
and boreability index of the rock mass.
In a more recent study by the authors (Hassanpour, 2009, 2010; 3. TBM field performance database
Hassanpour et al., 2009, 2010), based on data obtained from main
tunneling projects in Iran and investigating relationships between In this study data on geological and ground conditions, TBM
rock mass properties and actual machine performance, some new operational parameters and machine performance represented by
empirical equations have been proposed for estimation of TBM rate of penetration were collected during pre-construction and
performance in the given ground conditions. construction phases. The data were arranged in a special database
The above models, of course, have their advantages and disad- including 158 tunnel sections of four selected projects (Table 1)
vantages because of their origin and background. Some of them where the ground conditions and machine performance informa-
like original CSM model don’t consider the main influencing tion were valid and could be verified. The data sets comprised
parameters and some of them like NTNU model require special two main categories. The first category included machine perfor-
experiments originated from the drilling. These tests are not com- mance parameters like net boring time, length of mined section
monly available outside Norway. Also some of the models like QTBM and also the average of machine operational parameters (thrust,
are too complicated. QTBM model originates from Q system and in- RPM, power and applied torque) throughout the section. These
cludes too many parameters for practical application. In addition, parameters were obtained from the daily operating records and
some parameters are overlapped in this model (Gong and Zhao, the TBM data logger. Also the most important performance param-
2009). On the other hand growth of TBM manufacturing technol- eters including average rate of penetration (ROP), penetration per
ogy and existence of some shortcomings in the prediction models revolution (P), Field Penetration Index (FPI, Tarkoy and Marconi,
have made it necessary to perform more research on the develop- 1991), and specific energy (SE) have been calculated using formu-
ment of the new models. lae (1)–(4) as listed below:
In this study, compiled field data obtained from three main tun-
neling projects in Iran (SCE Company, 2004, 2006, 2008) as well as Lb
ROP ¼ ð1Þ
the Manapouri tunnel project recently completed in New Zealand tb
(URS Company, 2003), were used to establish a new concept for
rock mass boreability classification and a more general model for ROP 1000
P¼ ð2Þ
TBM performance estimation. RPM 60
Fn
2. Description of the projects used for this study FPI ¼ ð3Þ
P
For developing a more accurate TBM performance prediction 200 NTBM rmc F r
model that can be applied in different geological conditions, data SE ¼ ð4Þ
3 dTBM P
from different projects with different rock mass conditions were
collected and compiled in a database. As mentioned above, three where ROP is rate of penetration (m/h), Lb is boring length (m), tb is
long water conveyance tunnels recently constructed in different boring time (h), P is cutter penetration in each cutterhead
Table 1
Main characteristics of tunneling projects.
No Project Tunnel length Available data TBM type and manufacturer Construction TBM diameter
(km) (km) period (m)
1 Karaj water conveyance tunnel, Lot 1 (Iran) 15.9 15.9 Double shield (Herrenknecht) 2006–2009 4.65
2 Ghomrood water conveyance tunnel, Lots 2, 24.5 24.5 Double shield (Wirth) 2003–2009 4.525
3 & 4 (Iran)
3 Zagross water conveyance tunnel, Lot 2 26 5.3 Double shield (Herrenknecht) 2006–present 6.73
(Iran)
4 Manapouri second tailrace tunnel (New 10 9.7 Main beam open TBM (Robbins, 1997–2002 10.05
Zealand) Kvaerner-Markham)
Table 2
Geological characteristics of tunneling projects.
Fig. 1. Distribution curve and frequency histogram of rock mass and TBM performance parameters in the database.
Table 3
Summary results of determination of regression coefficients of different geological and geomechanical parameters with FPI.
revolution (mm/rev), RPM is cutterhead revolutions (rev/min), FPI is The second part of database or category of information included
Field Penetration Index (kN/cutter/mm/rev), Fn is cutter load or nor- some geological parameters such as intact rock properties (Com-
mal force (kN), SE is specific energy (MJ/m3), Fr is cutter rolling force pressive and tensile strength, quartz content, porosity), discontinu-
(kN), dTBM is TBM diameter (m), NTBM is number of cutters on the ity characteristics such as spacing, NTH class or fracture class
cutterhead and rmc is the weighted average cutter distance from developed in NTNU (Bruland, 1998), surface condition and also re-
center of rotation (m). sults of calculation of some rock mass parameters (like RQD, RMR,
J. Hassanpour et al. / Tunnelling and Underground Space Technology 26 (2011) 595–603 599
Table 4
(a) Variables and summary of the generated model for forward stepwise regression analysis; (b) significance of r-value and coefficients for generated model and (c) analysis of
variance for the significance of regression for generated model.
Fig. 4. Chart for estimating FPI based on rock mass properties driven from Eq. (10).
Table 5
Previous equations developed by authors for each tunnel project separately.
reasonable correlation between the performance and geological software (2002). After consideration of different combinations of
parameters. Using FPI and RMCI encompasses some of the most parameters, it appears that the best results could be obtained by
influential parameters including UCS (intact rock strength), RQD excluding the two parameters a and spacing. The results show
(degree of fracturing of the rock mass), average cutter thrust, good correlation between Ln(FPI) as response parameter and UCS
which is the most important TBM operational parameter and main and RQD as predictors, in a linear combination with a 95% confi-
controlling parameter for machine performance, and penetration, dence level. As a result, a new equation was introduced as follows:
which is the result of these interacting parameters. It should be
FPI ¼ expð0:008UCS þ 0:015RQD þ 1:384Þ ð10Þ
noted that in calculating FPI, machine penetration or P (mm/rev)
is used and that allows for scaling the results to any size machine As shown in Table 4, the regression coefficient (R2) for this equation
using the cutterhead rotational speed, or RPM. Obviously, this is 78.5%. It indicates that the above regression model explains 78.5%
equation can be a useful and comprehensive tool for predicting of the total variance of the 158 datasets. A simple t-test and F-test
ROP for hard rock TBMs. analysis of the results indicates that the correlations are real and
As mentioned before, in this study, multi-variable regression the coefficients are true (Table 4).
analysis was also used to find an empirical equation to relate FPI To evaluate accuracy of the model, the measured and calculated
as a function of machine performance parameters to geological values of FPI are compared in Fig. 3. As shown, most of the pre-
parameters. For this purpose, four rock mass properties including dicted values of FPI especially when FPI < 25 kN/cutter/mm/rev
UCS, joint spacing, RQD, and a (angle between tunnel axis and dis- are close to actual values.
continuity) were used as independent variables and the recorded To facilitate using of Eq. (10), a FPI prediction chart is also devel-
FPIs were treated as dependent variable. Influence of each variable oped and presented as Fig. 4. This chart can be used for quick esti-
on the FPI was evaluated using forward stepwise regression ana- mation of range of values for FPI in grounds with different rock
lyzes. Statistical analyzes were performed by Version 11.5 of SPSS strength and rock quality. As shown in this chart, different rock
J. Hassanpour et al. / Tunnelling and Underground Space Technology 26 (2011) 595–603 601
Table 6
Summary of ground conditions for various boreability classes.
Boreability FPI range Rock mass Stability condition TBM excavatability (relative Example
class (kN/mm/ boreability difficulty of ground for TBM use)
rev)
B-0 >70 Tough Completely stable Tough Very strong and massive quartzitic veins , intrusive and
metamorphic rocks
B-I 40–70 Fair-tough Stable Fair Massive igneous and metamorphic rocks
B-II 25–40 Good-fair Minor instabilities Good Blocky and jointed Tuffs, Tuffites, Limestones
B-III 15–25 Good Only local structural Very good Alternations of Sandstones, limestones and Shales
instabilities
B-IV 7–15 Very good Some major Good Alternations of thin bedded Shale and Sandstone layers
instabilities
B-V <7 Excellent Collapse, gripper May be problematic Highly foliated and schistose metamorphic rocks (Slate,
problems, squeeze, Phyllite, Graphite schist), Shale, Marlstone, thick fault zones
etc.
Fig. 7. Rock mass boreability prediction chart with actual FPI range for different geological units in the database.
Table 7
Example of machine performance estimation using developed model.
Rock mass A B C D E F
Rock mass boreability class B-0 B-I B-II B-III B-IV B-V
UCS (MPa) 210 160 100 75 50 30
RQD (%) 100 100 80 60 40 20
A: Theoretical machine performance
Theoretical advance rate
FPI (kN/cutter/mm/rev) (Fig. 3) 96 64 29 18 11 7
Assumed thrust (kN/cutter) 200
Assumed RPM (rev/min) 7
P (mm/rev) 2.08 3.11 6.78 11.18 18.44 29.20
ROP (m/h) 0.88 1.31 2.85 4.70 7.74 12.27
Assumed U (%) 25
AR (m/day) 5.3 7.8 17.1 28.2 46.5 73.6
B: Practical machine performance
Actual advance rate
FPI (kN/cutter/mm/rev) (Fig. 3) 96 64 29 18 11 7
Assumed thrust (kN/cutter) 220 180 180 180 80 50
Assumed RPM (rev/min) 7 7 7 7 4 3
P (mm/rev) 2.29 2.80 6.10 10.06 7.37 7.14
ROP (m/h) 0.96 1.18 2.56 4.23 1.77 1.29
Assumed U (%) 25 5
AR (m/day) 5.8 7.1 15.4 25.4 2.1 1.5
J. Hassanpour et al. / Tunnelling and Underground Space Technology 26 (2011) 595–603 603
stability issues and due to some limitations of machine and back- phic rocks and fractured rock masses. This model must be applied
up system such as capacity of conveyor belt. In general, more com- with caution in highly fractured rock masses (or fault zones) and
petent and stronger rocks indicated in classes B-0 and B-I coincide water sensitive rocks like marlstones and mudstones.
with higher utilization rate due to minimal ground support
requirement and related stoppages and downtime. On the opposite Acknowledgments
side of the scale, lower utilization rate is often experienced in
unstable grounds typical of class B-IV and B-V which reduces the Authors wish to thank SCE Company, especially our colleagues
utilization rate drastically and causes higher downtime and delays. in the Tunneling Division for their help in the collection of required
In addition, in softer grounds, operator reduces cutterhead thrust data.
to prevent cutterhead jamming due to high torque, minimizing
the chance of face collapse, and reducing the possibility of over- References
loading of conveyor belt. This simply means that same levels of ap-
plied cutter load are not used in soft ground (as it applies to Barton, N., 2000. TBM Tunneling In Jointed and Faulted Rock. Balkema, Rotterdam,
173.
stronger rocks) and by default the use of lower applied load will Barton, N., 1999. TBM performance estimation in rock using QTBM. Tunnels and
ease off the extremely high rates of penetration. Therefore, in part Tunneling International 31 (9), 30–33.
B of Table 7, by considering more practical values for utilization Bilgin, N., Kuzu, C., Eskikaya, S., 1997. Cutting performance of rock hammers and
roadheaders in Istanbul metro drivages. In: Golser, J., et al. (Eds.), Proceedings of
factor, thrust and RPM in such grounds, more realistic values for World Tunnel Congress 97 and 23th World Assembly of ITA: Tunnels for People,
penetration rate and advance rate of machine can be estimated. Vienna, Austria, Balkema, Rotterdam, pp. 455–460.
By reducing utilization rate, thrust and RPM to 5%, 80 kN/cutter Blindheim, O.T., 1979. Boreability Predictions for Tunneling. Ph.D. Thesis,
Department of Geological Engineering, The Norwegian Institute of Technology,
and 4 rpm in class B-IV and to 5%, 50 kN/cutter and 3 rpm in class p. 406.
B-V, estimated penetration rate and advance rate, have reduced Bruland, A., 1998. Hard Rock Tunnel Boring. Ph.D. Thesis, vol. 1–10, Norwegian
considerably. This is more in line with the practical application University of Science and Technology (NTNU), Trondheim, Norway.
Gong, Q.M., Zhao, J., 2009. Development of a rock mass characteristics model for
of TBMs in the field.
TBM penetration rate prediction. International Journal of Rock Mechanics and
However it is concluded that highest values of advance rate can Mining Science 46 (1), 8–18.
be achieved in rock masses categorized as classes B-II and B-III. In Hassanpour, J., Rostami, J., Khamehchiyan, M., Bruland, A., 2009. Development new
equations for performance prediction. Geo Mechanics and Geoengineering: An
such grounds, combined conditions of rock mass boreability and
International Journal 4 (4), 287–297.
stability of the surrounding rock are in the optimum condition Hassanpour, J., 2009. Investigation of the Effect of Engineering Geological
for excavating the tunnel by TBM. Parameters on TBM Performance and Modifications to Existing Prediction
Models. Ph.D. Thesis, Tarbiat Modares University, Tehran, Iran.
Hassanpour, J., Rostami, J., Khamehchiyan, M., Bruland, A., Tavakoli, H.R., 2010. TBM
6. Conclusion performance analysis in pyroclastic rocks, a case history of Karaj Water
Conveyance Tunnel (KWCT). Journal of Rock mechanics and Rock Engineering 4,
427–445.
Rock mass boreability depends on a number of influencing Hassanpour, J., 2010. Analysis of actual TBM performance in Ghomrood project,
parameters including intact rock/rock mass properties, machine Bulletin of Iranian tunneling Association, No 9 (article in Persian).
specifications and operational parameters. In this paper a simple Hoek, E., 2007. Rock Engineering, Course notes by Evert Hoek, p. 339, http://
www.Rocscience.com.
model is proposed to evaluate rock mass boreability and TBM per- Ramezanzadeh, A., 2005. Performance analysis and development of new models for
formance range. In the developed model, machine performance has performance prediction of hard rock TBMs in rock mass, Ph.D. Thesis, INSA,
been related to two main rock properties (UCS and RQD) and two Lyon, France, p. 333.
Ribacchi, R., Lembo-Fazio, A., 2005. Influence of rock mass parameters on the
operational parameters (average cutterhead thrust and RPM).
performance of a TBM in a Gneissic formation (Varzo tunnel). Rock Mechanics
These Input parameters of the model are typical parameters usu- and Rock Engineering 38 (2), 105–127.
ally are available even in the preliminary stages of the tunnel de- Rostami, J., Ozdemir, L., 1993. A new model for performance prediction of hard rock
TBM, In: Bowerman, L.D., et al. (Eds.), Proceedings of RETC, Boston, MA, pp. 793–
sign and planning. So, proposed model can be applied as a useful
809.
tool for quick estimation of TBM performance in projects with dif- Rostami, J., 1997. Development of a Force Estimation Model for Rock Fragmentation
ferent geological conditions and machine diameters. with Disc Cutters Through Theoretical Modeling and Physical Measurement of
This paper also introduces a new boreability classification based Crushed Zone Pressure. Ph. D. Thesis, Colorado School of Mines, Golden,
Colorado, USA, P. 249.
on rock masses characteristics to allow for prediction of FPI values. Sapigni, M., Berti, M., Behtaz, E., Busillo, A., Cardone, G., 2002. TBM performance
Various ground conditions are categorized in six different classes estimation using rock mass classification. International Journal of Rock
from B-0 as tough boring ground to B-V as easy ground for boring. Mechanics and Mining Sciences and Geomechanics Abstracts 39,
771–788.
Combining stability conditions which controls machine utilization SCE Company, 2004. Geological and Engineering Geological Report for Ghomrood
and boreability characteristics of different rock types and ground Water Conveyance Tunnel Project (Lots 3 & 4), Unpublished report.
condition allows for development of a new concept for classifica- SCE Company, 2006. Geological and Engineering Geological Report for Karaj Water
Conveyance Tunnel Project (Lot1), unpublished report.
tion of TBM excavatability or ‘‘Relative difficulty of ground for SCE Company, 2008. Geological and Engineering Geological Report for Zagross
TBM application’’. As mentioned, required FPI for boring the rock Water Conveyance Tunnel Project (Lot 2). Unpublished report.
mass, decreases from B-0 to B-V class. It means that from B-0 to SPSS Software, 2002. Statistical Package, version 11.5, SPSS Inc., Chicago, IL, USA,
<http://www.spss.com>.
B-V class, less cutterhead thrust is required to achieve a given rate
Tarkoy, P.J., Marconi, M., 1991. Difficult rock comminution and associated
of penetration. It is also concluded that rock masses with class B-II geological conditions. In: Tunneling 91, Sixth International Symposium, 14–
and B-III due to optimum conditions of stability and boreability are 18 April 1991, Hovotel, London, England, Institute of Mining and Metallurgy,
Elsevier Applied Science, London, pp. 195–207.
most favorite grounds for TBM application.
URS Company, 2003. Manapouri Power Station Second Tailrace Tunnel Engineering
The Developed empirical model is based on analyzing data ob- Geological Construction Report, Prepared for Meridian Energy Ltd., Unpublished
tained from four tunneling projects with total length of 55 km and report.
boring diameter of 4.5–10 m which have been excavated in differ- Yagiz, S., 2002. Development of Rock Fracture and Brittleness Indices to Quantify
the Effects of Rock Mass Features and Toughness in the CSM Model Basic
ent geological conditions. So, it can be considered a TBM perfor- Penetration for Hard Rock Tunneling Machines. Ph.D. Thesis, Department of
mance prediction model which is applicable in a wide range of Mining and Earth Systems Engineering, Colorado School of Mines, Golden,
geological conditions including layered and jointed sedimentary Colorado, USA, p. 289.
Yagiz, S., 2007. Utilizing rock mass properties for predicting TBM performance in
rocks, blocky – jointed pyroclastic and carbonate rocks, foliated hard rock condition. Tunnelling and Underground Space Technology 23 (3),
and schistose metamorphic rocks, massive igneous and metamor- 326–339.