Contribution by James Nicholas Shirlaw and Storer Boone
Nematollahi and Dias (2022) present a case study and numerical analysis of tunnelling based on the Line D subway extension in Lyon, constructed nearly 30 years ago. Previous studies on this tunnelling include those by Kastner et al. (1996), Bemnebarek et al. (2000) and Dias and Kastner (2013).
The title and statements within the paper refer to use of earth pressure balance (EPB) tunnel-boring machines (TBMs). There are also, confusingly, references to the same tunnels being driven by a slurry TBM. The density of slurry in the excavation chamber given on page 7 of the paper, 1224 kg/m3, would be suitable for bentonite slurry containing cuttings, but too low for typical EPB spoil. Earlier studies show that Line D extension tunnelling was by slurry TBM, specifically chosen because of the ground conditions (Bemnebarek et al., 2000; Kastner et al., 1996). The internal mechanics of slurry and EPB TBMs are significantly different, as is the mechanism of ground support (Anagnostou and Kovari, 1996) and variability of support pressure. Field monitoring data in the paper are thus for a different method of tunnelling than stated in the paper's title. The basis of modelling refers to EPB TBM tunnelling, so the comparison made between monitoring and modelling results appears to be for different tunnelling methods.
Monitoring data in the paper are for section P1–S1 of the Line D extension, close to the starting shaft at Vaise station, confirmed by the identifiers used for the five extensometers (EX 41, 13, 15, 11 and 16) shown in Figure 2 of the original paper and the same identifiers used in Figure 2 of the paper by Kastner et al. (1996). Figure 2 is the only figure in the paper under discussion that shows the location of extensometers relative to the tunnels. However, the five extensometers in Figure 2 are not referred to either in the text or in the other figures. The text refers to five extensometers EX 1, 2, 3, 4 and 5; some of these are also referenced in Figures 3, 12 and 16. The reader is therefore left uncertain as to whether:
- (a)
Figure 2 is irrelevant to the paper, as the work refers to the results from five different extensometers, for which no plan or section is given, or
- (b)
the authors chose to relabel the five extensometers shown in Figure 2 without informing the reader.
In the conclusions, the authors state on page 12 that ‘there are some differences in the water level, which could be due to the consideration of drained conditions for the clay layers’. Other than the initial water level provided in Figures 2 and 5, the figures show data for neither monitored nor modelled water level or pore pressure changes. Should the statement read: ‘there are some differences at the water level’, which would be consistent with the earlier discussion on horizontal movements in Section 4.3, rather than introducing new material in the conclusions?
One aspect of monitoring data discussed in detail by Kastner et al. (1996) was development of consolidation settlements over the tunnel at a significant distance behind the face of the TBM, starting 20 m after the shield passage. Table 3 in the paper by Kastner et al. (1996) indicated that consolidation settlement (phase 4) contributed about half of the total settlement at the ground surface at section P1–S1. Kastner et al. (1996) suggested that phase 4 settlement was due to consolidation of the ground and tail void grout. Dias and Kastner (2013: p. 19) comment on data from the same project and phase of settlement that the ‘rate of these settlements decreases significantly after 60 days, but there is still a slight evolution after one year’. Nematollahi and Dias (2022) state in their conclusions on page 12 that ‘The PH constitutive model can replicate the real soil behaviour during the tunnelling procedure’, based on comparing results from their modelling and field monitoring. However, none of the graphs presented shows the time or distance related development of settlement beyond about 14 m after passage of the TBM face. For real soil behaviour to be replicated, it would be necessary to show that modelling reasonably matched the observed settlement evolution and not just the movement over the TBM and the final settlement profile.
The authors state that the plastic hardening (PH) constitutive model was used to simulate stress–strain responses of the soils, but do not address the following issues.
- (a)
The PH model depends upon drained conditions prevailing during shear. The authors based their selection of drained behaviour on ‘the significant permeability values of the layers in the cross-section (10−3 to 10−5 m/s)’ (Nematollahi and Dias, 2022: p. 5). Figure 2 illustrates three soil layers within, and immediately above and below, the tunnel face: ‘gray clay’, ‘gray sand’ and ‘purple clay’. The layer of grey sand contained sufficient fines for Kastner et al. (1996), in their Table 1, to provide Atterberg limits. The soil descriptions of Nematollahi and Dias (2022) are inconsistent with their stated permeability range, but would suggest values orders of magnitude lower, inconsistent with fully drained behaviour. Lower values of permeability would be consistent with observed time-dependent settlement reported by Kastner et al. (1996). The paper provides no discussion regarding the influence or modelling of pore water pressures during construction or the simulated triaxial tests (including dissipation and tunnelling rate effects).
- (b)
The simulated triaxial tests appear to correspond well to laboratory data if the meaning of ‘exp’ noted in Figure 6 can be correctly assumed to reference experimental or laboratory results. The authors provide no discussion regarding selection of these laboratory results compared to natural variability, other available data or where these test samples were obtained compared to the instrumented tunnel section.
Figures in this paper comparing modelling results to field measurement data imply a remarkable accuracy and precision of the modelling to a few millimetres (or less in Figures 12 and 13). Given the nature of concerns discussed above, the reader is left to question whether this paper represents fortuitous results of off-setting assumptions or fitting to known field results.
