Tunnel boring machine (TBM) penetration tests entail incrementally increasing TBM thrust from full stop to maximum speed and recording the penetration rate either at set thrust levels or at set times. TBM penetration test data can be analysed by plotting the penetration rate (distance/revolution) against the net cutter thrust (force per cutter) over the full range of thrust levels in the test, called the penetration-thrust graph. This research shows that the transition from excavation dominated by grinding to excavation dominated by chipping can be observed in penetration-thrust graphs. Correlating penetration test data to the geological and geomechanical characteristics of rock masses through which a penetration test is conducted provides the ability to reveal the efficiency of the chipping process in response to changing geological conditions. By analysing penetration test data from projects in the Swiss Alps, and published data from Singapore, this research shows that the strength of the rock is an important control on how much net cutter thrust is required to transition from grinding to chipping. It also shows that the geological characteristics of a rock will determine how efficiently chipping occurs once it has begun. In particular, geological characteristics that lead to efficient fracture propagation, such as fabric and mica content, will lead to efficient chipping. These findings will enable a better correlation between TBM performance and geological conditions for use in TBM design, as a basis for contractual payments where penetration rate dominates the excavation cycle and in further academic investigations into the TBM excavation process.
During hard rock tunnel boring machine excavation, a cutter first creates a crushed zone at the cutter-rock interface and the stresses from the thrust of the cutter are transmitted through this crushed zone into the adjacent undamaged rock (Figure 1). The induced stresses and dilation within the crushed zone cause extensile fracturing of the rock away from the crushed zone. Eventually, fractures generated by subsequent cutter passes extend either to the rock surface or to fractures propagating from adjoining kerfs and coalesce to form chips. This occurs at different cutter thrust magnitudes for different rock types. If the cutter thrust necessary for tensile fracture propagation is not achieved, due to excessively high cutter thrust requirements or an underpowered TBM, then only grinding at the crushed zone occurs. Grinding produces fines, rather than chips, leading to much lower penetration rate. Chipping is a more efficient excavation process because generating chips through tensile fracturing is much more efficient than the formation of fines in the crushed zone (Teale, 1964; Snowdon et al., 1982; Bruland, 1998; Gertsch et al., 2007; Yin et al., 2014). The formation of chips by the chipping process is, therefore, critical for achieving high penetration rates.
Figure 1: Fracture and crushed zone creation at the cutter-rock interface during TBM excavation (modified from Villeneuve, 2008).
This research demonstrates what TBM penetration testing can show about the excavation process. Based on work by Villeneuve (2008, 2017) and Frenzel et al. (2012), penetration tests are defined and index values are demonstrated for TBM performance analysis. Geological characteristics are linked to the chipping process identified with the penetration tests. TBM operational data and results from penetration tests are analysed to provide feedback about whether excavation is occurring efficiently through chipping or inefficiently through grinding.
2 PENETRATION TESTS
Villeneuve (2017) performed a total of 16 penetration tests in three different rock units in the Swiss Alps: schist (Figure 2), granite (Figure 3) and gneiss (Figure 4). Figure 4 also includes data for marble and granite from Yin et al. (2014) and Gong et al. (2007). The tests in the Swiss Alps were undertaken using three Herrenknecht hard rock gripper TBMs with a diameter range of 8.83-9.58 m. All TBMs utilised 432 mm diameter cutters with 90 mm spacing on centre. During normal TBM start-up only a few data points at low penetration rates were recorded by the data acquisition system (DAS) due to the sampling interval (typically 1/10 s). In order to capture sufficient data through the full range of penetration rates, penetration tests were adopted (Villeneuve, 2008; later described in detail in Frenzel et al., 2012), and conducted by gradually increasing the TBM thrust from full stop to the maximum thrust over a period of 8-10 min. The cutterhead rotational speed (RPM) was kept constant during these tests, typically ranging from 5.5 rpm to 6.2 rpm, and was selected based on the face condition (i.e. it would be higher in stable face conditions than in blocky face conditions). Depending on the operator, RPM and rock type, the length of tunnel tested is approximately 30-200 mm.
The penetration rate (mm/rev) was used in Villeneuve (2017), rather than speed (mm/min), because this removes the effect of RPM and allows comparison of test results from different strokes. The thrust value obtained from the DAS is gross thrust, which is the amount of force exerted by the thrust pistons. This thrust incorporates friction on the TBM head, which is independent of cutting processes occurring at the tunnel face. The net cutter thrust is used, which is the gross thrust minus the frictional losses, divided by the number of cutters, to allow comparison of test results from different locations and different TBMs. The friction contribution to gross thrust is estimated by averaging the gross thrust required to reverse and advance the TBM cutterhead (i.e. moving the cutterhead when it is not touching the rock at the face, usually during cutter changes). Gross thrust also includes impacts of TBM stiffness and losses in the hydraulic systems, but these should remain constant for any TBM.
Villeneuve (2008) showed that penetration curves highlight (Figure 5): (1) the minimum thrust required to begin advancing the TBM; (2) initial penetration behaviour dominated by grinding (creating fines); (3) the change from grinding to chipping (creating chips), called the critical thrust (Robbins, 1970), located at the inflection point; (4) the chipping efficiency, represented by the slope of the line past the inflection point; and (5) the point of steady-state penetration.
The steady-state penetration rate is limited by TBM design parameters, which define performance limits. The penetration limit is a function of muck conveyance, bucket design, cutter wear and maximum head revolution speed (Frenzel et al., 2012). The torque limit provides the transition from the penetration limit to the maximum thrust (Frenzel et al., 2008), which is a function of the rotational speed and is controlled by the maximum torque capacity. The thrust limit is controlled by cutter type, maximum thrust capacity and TBM head design. The penetration limit for the TBMs used in this investigation is ~11 mm/rev (Villeneuve, 2017). The torque limit for the TBMs in this study is ~30% of the maximum torque capacity (Villeneuve, 2017). The maximum net thrust on 432 mm cutters is ~250-267 kN (Frenzel et al., 2008; Maidl et al., 2008).
Figure 2: Penetration test data for schist (modified from Villeneuve, 2008).
Figure 3: Penetration test data for granite (modified from Villeneuve, 2008).
Figure 4: Penetration test data for (a) gneiss, and (b) two marble samples (HT03 and DT) tested in Yin et al. (2014) and one granite sample (T05) from Gong et al. (2007) (modified from Villeneuve, 2008).
Figure 5: Penetration test data showing the key components of the curve (from Villeneuve, 2017).
3 PENETRATION TEST ANALYSIS
The minimum thrust and grinding portion of the penetration curve will depend on the resistance of the rock to crushing (Villeneuve, 2017). The critical thrust point and the slope of the curve during chipping (as discussed in Samuel and Seow (1984), Zhang et al. (2003) and Gehring (2009)), are related to rock strength and brittleness, mineralogy and fabric (Villeneuve et al., 2007, 2012; Villeneuve, 2008), and stress at the tunnel face (Yin et al., 2014). The location of the critical thrust and the slope and length of the curve beyond the critical thrust point are representative of the 100-200 mm thickness of rock over which the test is performed. Performing the test in different lithologies will produce different curves, which can be related back to the geological conditions.
The UCS values for the schist and granite in which the penetration tests were conducted show a strong relationship with critical thrust (Figure 6). The variance in these data likely results from fractures (pre-existing or stress-induced) in the face and variability at the metre scale in the rock strength. This shows that, while UCS can be used to identify rocks that are at risk of poor excavation performance, penetration tests are required to identify the actual point at which chipping occurs.