Vous êtes sur la page 1sur 8

ISRM International Symposium 2008

5th Asian Rock Mechanics Symposium (ARMS5), 24-26 November 2008 Tehran, Iran

RECOMMENDED ROCK TESTING METHODS FOR PREDICTING TBM PERFORMANCE: FOCUS ON THE CSM AND NTNU MODELS
S. YAGIZ1, J. ROSTAMI2, L. OZDEMIR3
1

Pamukkale University, Geological Engineering Department, Denizli, Turkey (e-mail of corresponding author: syagiz@pau.edu.tr)

Penn State University, Energy and Mineral Engineering, University park, PA, USA 3 Earth Mechanics Institute of Colorado School of Mines, Golden, CO USA

Abstract
The performance and efficiency of tunnel boring machines (TBM) depend on the rock properties and geological structures, and machine specifications and operational parameters. Several predictor models are currently available for estimation of TBM performance in rock masses. Models developed by the Colorado School of Mines (CSM) and the Norwegian University of Science and Technology (NTNU) are two most common predictors for estimating the TBM performance. Each model utilizes different input parameter for predicting the TBM penetration rate. Also they use different rock testing methods to obtain related rock properties. The CSM Model is based on force equilibrium method and uses intact rock properties including uniaxial compressive strength (UCS), Brazilian indirect tensile strength (BTS) and abrasiveness of rock. The NTNU Model is empirical method and focuses on geological structures such as joint, plane of weakness, bedding as well as porosity, brittleness, and abrasiveness of rock. Also, the Modified CSM Model that were developed based on comparison of the predictions of the base CSM Model and field data has been introduced by some researchers. The new model includes rock brittleness and the rock fracture properties as adjustment factors into the existing model. These approaches are briefly discussed in this paper. This paper will compares these models and their input parameters, their input variables as well as predicted performance results. Further relevant rock properties and related laboratory testing methods will be highlighted. Keywords: Tunneling, Tunnel Boring Rock testing; TBMs; CSM; NTNU, performance prediction.

1. Introduction One of the most important parameters for predicting the TBM penetration rate is engineering properties of the rock mass. To predict TBM performance, several models have been developed by various researchers [114].The two main group of models include empirical and semi-theoretical approaches. The interrelationship between cutter wear, machine operation, continuous mucking, and support installations requires an evaluation of many factors affecting TBM performance. All mechanical rock-cutting tools share the same principles; hence, many efforts have been made to develop performance prediction models and theories offering explanations into the forcepenetration behavior of rocks [9, 14-20]. The rock cutting process deals with the indentation of a rock surface by a cutting tool. In the case of disc cutters, which are the most common cutting tool on hard rock TBMs, the indentation is followed by a forward movement, leaving

behind it a groove and fractured and crushed rock [21]. The uniaxial compressive strength (UCS) and Brazilian tensile strength (BTS) of the rock are the most common rock properties in various mining and civil construction projects. These tests are often used as input parameters in TBM performance prediction models. However, the rock strength alone is not enough to predict penetration rate of the TBM, where the fractured faulted rock masses are to be encountered. Therefore, various rock tests and data evaluation techniques have been introduced to account for rock mass properties [9, 13, 21-24]. The most common models to predicting TBM penetration rate are the Colorado School of Mines (CSM) and the Norwegian University of Science and Technology (NTNU) performance models [1, 3]. Since their original developments, the models have been improved and updated using available additional dataset to improve their accuracy [4, 5, 9, 11 and 12]. This paper will review the rock

1523

properties that affect TBM performance and compare the rock testing methods performed for the original CSM, the Modified CSM, and the NTNU models. An example tunnel will be discussed to compare the result of these models. 2. Background In general, the TBM penetration rate and cutter wear are the important issues for TBM application in tunnels, and they are a function of rock strength, hardness, toughness, brittleness, anisotropy and abrasivity. Any rock directional properties caused by existing foliation, faults, bedding, grain alignment or other defects, and localized features may significantly influence the crack initiation and development in the rock and the resultant of rock chipping process. The machine specifications, such as thrust and power are the key to providing sufficient amount of forces and torque to support the excavation operation. Machine thrust should provide enough force for the tools to efficiently penetrate into the rock surface. Also, the cutterhead torque and power must be sufficient to ensure rotation of the head at the required penetration rate and overcome the rolling force resistance of the cutters. As the cutter penetrates the rock, a crushed zone is formed beneath the cutters due to extremely high stress generated in the rock by the concentrated forces at the cutter tip. The tensile stresses generated by the crushed zone causes radial crack to initiate and propagate in the rock surface. The cracks continue to propagation till the tensile stress at their tip falls below the tensile strength of rock (Figure 1). If the stress developed in the crushed zone is sufficiently high, one or more crack extended far enough to reach one of the tensile crack developed from adjacent cut, causing rock failure in the form of a chip [1, 4, 9]. The laboratory or field testing should provide relevant information on the rock properties impacting this process.
Figure 1. Rock Breaking and Failure Mechanism

3. Suggested Laboratory Tests for TBM Performance Prediction As discussed before, the most important rock properties for investigating the TBM penetration rate are strength, hardness, brittleness, toughness, and also rock discontinuity properties encountered during the tunneling. Thus, different rock testing methods and evaluation techniques were introduced in the literature. In the original CSM model, UCS, BTS, and Cerchar abrasivity Index (CAI) of rock is used, while distance between plane of weakness (DPW), alpha angle (), the angle between plane of weakness and TBM driven direction, and punch penetration test (PSI) are the additional factors used in the Modified CSM model. Sievers J-value (SJ), rock brittleness (S20), abrasivity value (AV) , and porosity are the rock properties used input parameters for the NTNU model, together with rock mass properties including joint orientation and spacing encountered in the tunnel. Following is a brief description of each test method. 3.1. Rock Sampling One of the key factors for all TBM project planning and predicting performance is careful sampling of anticipated lithologies along the tunnel from surface and subsurface investigations. This is done to evaluate ground conditions at the site which can impact machine performance. The sample selection criteria should focus on the physical and mechanical properties of rocks after completing the geological mapping. The number of samples required depends on the complexity of the site

1524

geology, tunnel length and also intended rock test methods. 3.2. Rock Strength Tests (UCS and BTS) The uniaxial compressive strength (UCS) of rock is performed according to the ISRM [25] and ASTM [26, 27] suggested standards. The UCS is the most commonly measured rock property to evaluate the potential resistance of the rock against the indentation of the cutting tool into the rock surface. Thus, many of the developed models and equations use UCS as input parameters. However, to achieve the viable relation between the rate of penetration (ROP) and the UCS, the great attention must be paid to sample failure mode. The samples that fail along existing rock defects such as joints, bedding, and schistosity should be classified as structural failure. Where the sample failure is not controlled by any defects and occurred in an intact manner, the sample should be noted as having failed in a non-structural and be excluded. This is important since the structural failures do not represent the actual rock strength and should not be used for predicting TBM performance. It must be noted that many prediction models, rely on the UCS results, and any discrepancies in the testing may produce inaccurate results [28-29]. The Brazilian tensile strength (BTS) of the rock is conducted according to the ISRM [25] and ASTM [30] standards as well. The BTS is also used in boreability predictions together with the UCS. BTS is generally intended to provide an indication of rock toughness/brittleness from a viewpoint of crack propagation between adjacent cuts. Like the UCS testing, the failure of the rock under the tensile loading should be carefully observed to identify the type of failure. When the sample fails along the defect, the result must be noted in the dataset used for the prediction of TBM performance. The both rock strength parameters (UCS and BTS) are the main input variable for the CSM model. However, the NTNU model does not use these measurements in their prognosis.

Meanwhile, rock porosity measured according to the ISRM (1981) Suggested Testing Methods has been used as one of input parameters into the NTNU model but not into the CSM Model. 3.3. Rock Hardness Tests, (Sievers JValue) All the minerals in rocks have distinctive scratch hardness. The Mohs hardness scale, used for defining the hardness is divided into ten increments, ranging from talc to diamond. Besides, various rock testing methods including the Schmidt rebound hardness, Shore scleroscope hardness, Vickers hardness, Taber abrasion and Sievers J-value are also some measure of the rock hardness used in various applications. The CSM model or the Modified CSM model does not use any of these definitions of rock hardness as input parameter. On the other hand, the Sievers J-value that is a measure of rock surface hardness is utilized as input parameters to estimate drilling rate index (DRI) in the NTNU model (Figure 2). The Sievers J-value is measured by miniature drilling with a bit diameter of 8.5 mm. The depth of the hole after 200 revolutions, measured in 0.1mm, taken as Sievers J value for that test. The SJ value of the rock is the mean of 4-8 tests, depending on the consistency of the results. The precut surface of the sample must be parallel or perpendicular to the foliation of the rock; however, The Sievers J-value measured parallel to the foliation is used to compute the DRI that may be described as the brittleness value adjusted for the Sievers J-value in the NTNU model as shown in Figure 3 [3132].

1525

Figure 2. The Sievers J-Value Drill Test [31-32]

measure rock brittleness and toughness. Data evaluation techniques, usefulness and producers of the test were discussed in detail for various purposes by different researchers [9, 21, 23, 29, and 33]. The test results that is the fluctuation of the force-penetration profile obtained from the test is indirectly used to evaluate the toughness and brittleness of rock in the Modified CSM model by visual observation of the forcepenetration curve. Later on, the brittleness index (BI) was introduced as the slope of the obtained graph from the origin to maximum applied force. This index was developed and utilized as one of the quantitative intact rock properties together with the UCS and BTS of rock into the Modified CSM model as given in Figure 4a, b [9]. The NTNU brittleness test (S20) is used to define the drilling rate index (DRI) that is one of the inputs for the prognosis model. The test gives a measure of the ability of the rock to resist fracturing from repeated impacts [31-32]. The volume of the rock for testing corresponds to a 500 gram specimen obtained by sieving the crushed rock sample and range 16-11.2mm in size. The brittleness value equals the percentage of materials passing the 11.2mm mesh sieve after the aggregate has been crushed in the mortar. The S-20 value is the mean value of 3-4 repeated tests (Figure 5). 3.5. Rock Abrasiveness Tests (CAI and AVS) The Cerchar abrasivity index (CAI) test is used for measuring the abrasiveness of rock for estimation of cutter wear [9, 31 and 34]. The test is performed by scratching a freshly broken rock surface with a sharp pin of heat-treated alloy steel. The CAI could be computed as the average values of the abraded tip of the steel pin in 0.1 mm after one cm of the travel across the rock surface. The CAI value is the average of 10 measurements, 5 pins tested, diameter of flat measured in two diagonal directions, Cerchar Index is used in the CSM model to estimate cutter wear and life that is directly related to the resultant wear flats in the field.

Figure 3. Computation of the DRI in the Model [32]

3.4. Rock Brittleness Test (BI and S20) Punch penetration test is one of the methods to measurement of the rock brittleness index (BI). The brittleness is also one of the most effective and required rock properties for estimating the TBM rate of penetration. In general, the rock cutting efficiency of any mechanical tool improves with increasing brittleness exhibited by rocks. Thus, the brittleness is a highly desirable feature of the rock from boring point. Punch penetration test does not have any institutional standard, but is utilized according to industrial standard of practice. The test is conducted to the

1526

Figure 4a. Punch Penetration Test Set up [9, 33]


6.0 5.5 5.0 4.5 4.7 4.4 4.1 3.8 3.5 3.0 2.5 2.0 1.5 1.3 1.0 0.5 0.0 0 0.0 50 100 150 200 250 300 350 400 0.6 0.3 0.9 1.6 2.2 1.9 2.5 3.1 2.8 3.5 5.0 5.3

Figure 5. The NTNU Brittleness Test (S20) [31-32]

3.6. Petrography (Thin Section Analysis) A reliable determining the microscopic properties of rock can be obtained within thin section analysis that is the study of a transparent section of the rock under microscope. Type and content of hard minerals in rock (i.e. garnet, quartz etc); grain size, shape, elongation and orientation; grain suturing and interlocking; micro fractures and any other invisible micro feature of rock can be identified by using thin section analysis. Besides, X-ray diffraction (XRD), X-ray fluorescence (XRF) may be used for quantifying the chemical and semi quantitative determination of the mineral composition as necessary. The result of the tests can be used for investigating the mineralogical, micro-structural and textural composition of rock in the both Models. 4. Rock Mass Properties In rock mass, discontinuities, foliations, faults and beddings play a significant role in controlling fracture propagation between cuts, depending on the orientation of the rock discontinuity with respect to the direction of TBM advancement. The discontinuity properties greatly affecting the TBM penetration rate should be quantified to use for predictors. The NTNU model uses fracture properties, including that is the angle between plane of weakness and the TBM driven and joint spacing, which is the distance between planes of weakness (DPW),

Brittleness Index (BI)

4.0

Peak slope of the test, PSI (kips)

Figure 4b. Computation of the BI as Function of PunchPenetration Tests in the Modified CSM Model [9]

In the NTNU model, abrasive value (AV) that is defined as the weight loss in mg of a tungsten carbide test piece after 20 rotations (corresponding time of test is one minute) on the rock powder spread on a rotating plate. AV is used for estimating the bit wear index (BWI) and cutter life index (CLI) together with Sievers Jvalue. Further, abrasion value of steel (AVs) that is the weight loss in milligrams of the bit made of a hardened tool grad steel after 100 revolutions (corresponding time is 5 minutes) of the disc is utilized for estimating bit wear. Both AV and AVS tests are measured with same equipment [31-32]. The rock abrasiveness is the main property for estimating the cutter wear but not for predicting the penetration rate (ROP).

1527

as input parameters to estimating the rate of penetration [3, 5, 9, 22]. The rock mass fracture properties are not used in the CSM model directly; however, in the Modified CSM model Rock Fracture Index (RFI) were developed for accounting both alpha angle and DPW as adjustment factors to predict the TBM penetration rate [9]. Yagiz (2002) concluded that the rock mass fracture properties including the RFI together with the BI are the most affective parameters for estimating the TBM penetration rate in faulted and fractured rock mass condition (Figure 6).
3.0
90 o 80 o 70 o 60 o 50 o 40 o 30 o 20 o 10 o

2.0

1.0

0.0 0 5

10 15 20 25 30 35 40 45 50 55 60 65 70 75 80

-1.0

DPW (inches)

-2.0
0o

formulas for adjustment factors that was developed by using a different and more diverse set of field data; however, there are some similarities between the two systems [36]. The NTNU model is the empirical performance predictor are based on the historical field performance database that mostly includes tunnels in Norway. The model offers a set of empirical graphs and equations obtained from various regression analyses between rock properties, ground conditions, machine properties and the ROP. As result, while the NTNU model is able to incorporate rock mass properties and ground condition directly into its predictions. The CSM models are more flexible with machine design parameters (i.e. thrust, torque, power) and allow design modification for optimizing the layout and balance of cutterhead. Two predictors have been compared several times with the results being very close on many cases where the joints and discontinuities do not affect borability (Table 1).
Table 1. Comparison between the performance predictions for Yucca Mountain Welded tuff [32, 35].

Rock Fracture Index (RFI)

-3.0 Alpha 0 50 Degree 10 Degree 60 Degree 20 Degree 70 Degree 30 Degee 80 Degree 40 Degree 90 Degree

Figure 6. Computation of the RFI as Function of the Alpha Angle and DPW in the Modified CSM Model [9]

Standard TBM CSM ROP (mm/rev) ROP (m/hr) Cutter Life (m/cutter) 6.09 2.33 3.44 NTNU 5.94 2.28 5.26

H-Power TBM CSM 8.88 3.73 6.86 NTNU 7.89 3.31 9.48

5. Principles of the CSM and NTNU Performance Models The CSM model is based on estimation of the cutting forces acting on individual cutters and determines the overall thrust, torque and power requirement of the entire cutterhead for given rock properties and machine specifications. The estimated values of cutterload, thrust, and power are then compared to corresponding machine specifications to obtain maximum penetration rate by try and error. This provides the estimated penetration as function of machine and rock properties (UCS, BTS). Further, the Modified CSM model was developed bases on the original CSM Model concept by adding the rock brittleness (BI) and RFI as an adjustment factor to account for rock brittleness and rock mass parameters [9]. There are different set of

Hence, the Modified CSM and NTNU models obtain the results that quite agree in the faulted and fractured rock condition. 6. Conclusions The paper was provided an overview and interpretation of the rock tests for the CSM and NTNU models. New improvements in the CSM model are also briefly discussed by introducing the Modified CSM model and related adjustment factors. The both intact rock properties including strength, toughness, brittleness, hardness and also rock mass properties including joints,

1528

fractures, bedding, weakness zones and distance between plane of weakness are crucial parameters for the prediction of TBM performance in hard rock. Therefore, before selection of the TBM for a given tunneling project, the tunnel site should be carefully investigated by conducting laboratory rock testing and also site investigation and geological mapping. Both the CSM and NTNU as well as other research groups around the world continue to improve their models by adding more rock properties, rock testing methods and also more field data from past and current TBM cases in the databases. As results, all these effort would contribute to the development of more precise TBM performance prognosis models for future. Acknowledgments Partial grant provided by the Scientific Research Center of Pamukkale University (PAU-BAP) to attend the conference is acknowledged. References
[1] Ozdemir, L. 1977. Development of theoretical equations for predicting tunnel borability. Ph.D., Thesis, T-1969, CSM, Co USA [2] Tarkoy, P.J. 1975. Rock hardness index properties and geotechnical parameters for predicting TBM performance. PhD thesis, 325p. University of Illinois at Urbana Champaign [3] Blindheim, O.T. 1979. Boreablity predictions for tunneling. PhD thesis, Dept of Geological Engineering in the Norwegian University of Science and Technology , 406p. [4] Rostami, J. and L. Ozdemir. 1993. A New Model for Performance Prediction of Hard Rock TBM. In Bownerman and Monsees (Eds), In Proceedings of Rapid Excavation and Tunneling Conference, Chapter 50: p793-809 Boston, USA [5] Bruland, A. 1999. Hard rock tunnel boring advance rate and cutter wear. Norwegian Univ. of Science and Technology , 183p. Norway, [6] Barton, N. 2000. TBM tunneling in jointed and faulted rock. p173. Balkema, Brookfield [7] Nelson, P.P. and T.D. ORourke. 1983. Tunnel

boring machine performance in sedimentary rocks. Civil and Env. Engineering, Cornell University. No: 83-3, p438. Ithaca, NY [8] Nilson, B. and zdemir, L. 1993. Hard rock tunnel boring prediction and field performance. Chapter 52, Proceedings of Rapid Excavation and Tunneling Conference, pp833-852. [9] Yagiz, S. 2002. Development of rock fracture and brittleness indices to quantifying the effects of rock mass features & toughness in the CSM Model basic penetration for hard rock tunneling machines. Ph.D .Thesis. T-5605. pp.289. Dept of Mining and Earth Systems Engineering, Colorado School of Mines. Golden Co. USA [10] Yagiz S. 2003. A model for prediction of TBM performance in hard rock condition. Abstract. p227. 56th Geological Congress of Turkey, MTA General Directory, Ankara, Turkey [11] Yagiz, S. 2006a. Recent advancement in predicting TBM performance. 7th International Congress on Advances in Civil Engineering. 10p. in DVD. October 11-13 2006. Istanbul TR. [12] Yagiz, S. 2006b. A model for prediction of tunnel boring machine performance. Paper no. 383. Engineering geology for tomorrow`s cities. 10p in DVD. The 10th International Association of Engineering Geologists Congress. The Geological Society of London. U.K. [13] Yagiz, S. 2006c. TBM performance prediction based on rock properties. EUROCK2006. Multiphysics Coupling and Long Term Behavior in Rock Mechanics. (Editors) Cotthem, A., R. Charlier, J. Thimus, J. Tshibangu pp663-670. University of Liege. Belgium. [14] Yagiz, S. 2008a. Utilizing rock mass properties for predicting TBM performance in hard rock condition. Tunneling and Underground Space Technology. 2. p 326-339. [15] Spagni, M., M. Berti, E. Bethaz, A., Busillo, G. Cardone 2002. TBM performance estimation using rock mass classficaitions. Int Journal of Rock Mech and Min Sci. 39, 771-788. [16] Graham, P.C. 1976. Rock exploration for machine manufacturers. In exploration for rock engineering. Proc. of the Symp. Johannesburg,

1529

In:Bieniawski, Z.T.(Ed) Vol.1, pp173-180. [17] Farmer, I.W. and NH. Glossop. 1980. Mechanics of disc cutter penetration. Tunnels and Tunneling 12(6), 22-25 [18] Roxborough, F.F. and HR. Phillips. 1975. Rock excavation by disc cutter. Int. J. Rock Mech. Min. Science and Geomechanics, Abstracts V.12, p361-366. [19] Sanio, H.P. 1985. Prediction of the performance of disc cutters in anisotropy rocks. Int. Journal of Rock Mechanics and Mining Science, V.22, No.3, p153-161 [20] Sato, K., F. Gong, K. Itakura. 1991. Prediction of disc cutter performance using a circular rock cutting ring. Proceedings 1st International Mine Mechanization and Automation Symposium, June, CSM, Co, USA [21] Dollinger, G.L. and J.H. Handewith. C.D., Breeds. 1998. Use of punch tests for estimating TBM performance. Tunneling and Underground Space Technology, Vol.13, No.14, pp403-408. [22] Lislerud, A. 1988. Hard rock tunnel boring: prognosis and costs. Tunneling and Underground Space Tech. Vol.3(1), p9-17. [23] Szwedzicki, T. 1998. Draft ISRM Suggested method for determining the indentation hardness index of rock materials. Int. J. Rock Mech. Min. Sci., V 35. No.6, p831-835. [24] Yagiz, S. and L. Ozdemir 2001. Geotechnical parameters influencing the TBM performance in various rocks. Program with Abstracts. p79. 44th Annual Meeting of AEG. Eng. Geology for Construction. MO USA. [25] ISRM, 1981. Rock characterization testing and monitoring, ISRM Suggested Methods. International Society for Rock Mechanics. In: Brown, E.T. (Ed.) 167s. UK [26] ASTM, 1995. Standard practice for preparing rock core specimens and determining dimension and shape tolerances. American Society for Testing and Materials, D4543. [27] ASTM, 1995. Standard test method for unconfined compressive strength of intact rock core specimens. American Society for Testing and

Materials, Designation D2938. [28] Cigla, M., S. Yagiz, L. Ozdemir. 2001. Application of tunnel boring machines in underground mining development. 17th Int Mining Congress and Exhibition of Turkey. pp155-164. Ankara TR [29] Yagiz S, Ozdemir L, Rostami J. 2008c. Laboratory tests suggested for mechanical excavation in rock mass. IX. Regional Rock Mechanics Symposium. 10p. DEU, Izmir. [30] ASTM, 1995. Standard test method for splitting tensile strength of intact rock core specimens. American Society for Testing and Materials, Designation D3967. [31] Nilson, B. and L. zdemir. 1999. Recent developments in site investigation and testing for hard rock TBM projects. Proceedings of RETC. pp715-731 as chapter 39. [32] Bruland, A. and B. Nilsen. 1995. Tunneling performance estimation based on drillibility testing. International Congress on Rock Mechanics. pp123126. Tokyo Japan. [33] Yagiz, S. 2008b. Assessment of brittleness using rock strength and density with punch penetrationtest. DOI: 10.1016/j.tust.2008.04.002, Tunneling and Underground Space Technology. [34] CERCHAR, 1986. Centre d Etudes et Recherches de Charbonnages de France. The Cerchar abrasiveness index, 12 S. Verneuil Fr [35] Rostami, J. L. Ozdemir, B., Nilsen. 1996. Comparision between CSM and NTNU Hard rock TBM performance model. Proc. of Annual Technical Meeting of the Institute of Shaft and Drilling Technology, Las Vegas, USA. [36] Ramezanzadeh, A., Rostami, J., Kastner, R., 2005. Influence of rock mass properties on performance of hard tock TBMs. Rapid Excavation and Tunneling Conference, Seattle, WA, June 2729, 2005.

1530

Vous aimerez peut-être aussi