The instrumentation layout shown in Fig. 1 was designed specifically to monitor the subsurface ground responses induced by the construction of the westbound and eastbound tunnels. Installations for measuring pore water pressures include three boreholes containing multi-level VW piezometers, and one with a conventional standpipe piezometer. There are also four boreholes with combined total stress transducer-piezometer spade cells. Cross-sections A–A and B–B marked in Fig. 1 indicate the relative positions of the different types of piezometer sensors and the Crossrail tunnels and are shown in Figs 2(a) and 2(b).

Within Hyde Park, two piezometer boreholes (HP32 and HP33), with six piezometer sensors at various depths in each, were installed above and adjacent to the Crossrail eastbound tunnel. One spade cell (HP39) was installed 1·5 m directly above the same eastbound tunnel crown and three others (HP35, HP36 and HP37) either side of it, all being at tunnel axis level (Fig. 2(a)). Under the southern pavement of Bayswater Road, there is one multi-level piezometer borehole (HP34), containing six sensors, directly above the Crossrail westbound tunnel (at depths down to 2 m above crown level) and at a horizontal offset of 5 m from the existing Central Line westbound tunnel extrados (Fig. 2(b)).

The multi-level piezometers were installed in the boreholes using a fully grouted method where the VW sensors (manufactured by Soil Instruments Ltd), equipped with high-air-entry filters and having a measurement range of 300 kPa or 500 kPa, were fixed at pre-determined levels within the borehole before backfilling with a suitable cement–bentonite grout. Wan & Standing (2014a) discuss in detail the installation, including the selection of suitable grout mix proportions. Wan & Standing (2014b) present the background and pre-construction pore water pressures measured by these piezometers and the resulting steady-state pore water pressure profiles at the site. Good agreement between the measurements from the standpipe piezometer (HP31A) and the VW piezometers was demonstrated. Detailed analysis of the results also revealed the potential influence of claystones present in the LCF on the groundwater regime and the performance of the multi-level piezometers. It was established that the three bottom-most piezometer sensors in HP33 were hydraulically interconnected due to a high concentration of claystones over the relevant depth range. Wan & Standing (2014b) also present a reduced pore water pressure profile at the current site measured close to the existing Central Line tunnels (constructed with segmental cast-iron linings in the nineteenth century), suggesting that they have been draining the surrounding ground (LCF).

The spade cell type installed at the site (also manufactured by Soil Instruments Ltd) consists of a flat, rectangular (with a point at one end), spade-shaped oil chamber (formed by two 100 mm wide steel plates) connected to a VW pressure transducer with a measurement range of 2000 kPa. This measures total stresses acting on the pressure cell in the direction perpendicular to the spade surface. In-built at the top of one side of the spade is a porous filter, which is connected to a separate VW pressure transducer, so that the pore water pressure is measured independently at the same time. Each spade cell was installed by first drilling a borehole to a pre-determined depth (about 1 m above the target instrument depth). The spade cell, attached to a string of rods, was then lowered to the borehole base, before being pushed vertically to the target depth. The orientation of each spade cell was carefully adjusted so that the flat spade was parallel to the alignment of the eastbound Crossrail tunnel so as to make measurements of stress changes in the transverse direction. Wan & Standing (2014b) describe the installation procedure.

From the literature it is well established that pushed-in spade cells provide overestimates of ground stresses. Tedd & Charles (1981) investigated the measurement of in situ stress in London Clay by comparing horizontal stress measurements using spade cells with other in situ stress measuring instruments such as self-boring pressuremeters, whereas Tedd & Charles (1983) and Ryley & Carder (1995) compared the in situ overburden pressure with the vertical stress measured by spade cells installed horizontally, and found that the measured values need to be corrected by about half the undrained shear strength. Tedd et al. (1989) reviewed spade cell measurements from a number of sites in soft and stiff clays and found that generally the spade cells tend to over-read the actual total horizontal stress due to the complex local stresses on the spade surface induced by the installation pushing process. The over-read error based on all the above studies depends on the soil stiffness and can range widely from 0·3 to 2·0 times the undrained shear strength. It should be noted that the wide range of over-read factors determined from these studies can be attributed to a combination of errors associated with the ‘true readings’ that the spade cell measurements were compared with, and variability in the measurement of the undrained shear strength induced by different testing methods and sample dimensions. More recently, Richards et al. (2007) investigated the total vertical stress measured by horizontally installed spade cells under a box excavation in stiff Atherfield Clay, and reported that the over-read was about 0·35 times the undrained shear strength (determined from unconsolidated undrained tests on 100 mm dia. samples and standard penetration test measurements) compared with the actual overburden pressure. They also found that the spade cells under-read the change (reduction) of vertical stress upon the removal of overburden due to excavation, but concluded that the discrepancy was a result of shear stress mobilised between the ground and the perimeter wall (wall adhesion) resisting the ground heave, rather than a result of over-compliance of the spade cell (i.e. it being insufficiently stiff compared with the ground).

In general, the spade cell would either over-read or under-read the change of total stress depending on the relative cell/soil stiffness. For the type of spade cell used at the Hyde Park site, the cell stiffness is high compared even with the small strain stiffness of London Clay. Therefore the spade cells would tend to over-read the horizontal stress changes induced by the tunnel construction, although it is difficult to quantify the over-read accurately.

Having discussed the various limitations of using spade cells, it is also worth commenting that when monitoring the three-dimensional ground response to EPBM tunnelling, there is little ambiguity in measuring displacements, as individual devices (e.g. extensometers and inclinometers) are specifically designed to measure respective vertical and horizontal components. However, when measuring pore water pressure and total stress changes, a greater appreciation of the three-dimensional effects is needed to interpret the field data correctly.

The operation variables of the two EPBMs were recorded as they advanced. Wan et al. (2017a; Fig. 23) present the recorded face pressure, tail grout pressure and tail grout volume for both TBM1 and TBM2 (west- and eastbound construction, respectively) relating to when they were driven over about a 1 km distance beneath Hyde Park. In terms of rolling averages over ten lining rings, face pressures were maintained at about 200 kPa for both TBM1 and TBM2, while tail grout pressures were slightly more variable but broadly maintained at about 100 kPa for TBM1 and 200 kPa for TBM2. The in situ total overburden pressure at tunnel axis level under Hyde Park is about 650 kPa.

The EPBM face pressures and tail grout pressures are examined in more detail for when the TBM drives were in close proximity of where the piezometers were installed. Figs 4(a) and 4(b) show the EPBM pressures, based on rolling averages over three rings, within ±30 m of the piezometers and spade cells for TBM1 and TBM2, respectively. As both TBMs approached and passed the instruments, the face pressures were fairly consistent, ranging mostly between about 180 kPa and 220 kPa, whereas the tail grout pressures were more variable, ranging from 40 kPa to 240 kPa. As the tail grout pressures relate to the rear of the TBMs, for clarity, schematic representations of the TBMs in relation to the monitoring point are provided in Fig. 4. When comparing EPBM pressures with measured pore water pressure and total stress changes, the EPBM values shown in Table 1 are adopted for each instrument.

Fig. 4. Measured average face pressure and tail grout pressure of: (a) TBM1; (b) TBM2

Table 1. Average face pressure and tail grout pressure when the cutter-head or shield tail of both TBMs passed the instrument boreholes

Post-installation and steady-state pore water pressure and total horizontal stress measurements before the tunnel construction (period 1) are discussed by Wan & Standing (2014b) and Wan (2014), while this paper concentrates on the short-term responses observed during periods 2 and 4. The profiles of steady-state pore water pressure before the arrival of TBM1 as reported in Wan & Standing (2014b) are shown in Fig. 3. Note that before the tunnel construction, all piezometers were expected to measure greenfield ground pore water pressures except for HP34, which is located 5 m from the existing Central Line westbound tunnel. The greenfield measurements from HP32 (Fig. 3(a)) indicate a slightly under-drained steady-state piezometric profile within the LCF, while it is evident from the measurements from HP34 (Fig. 3(b)) that the steady-state pore water pressure near the existing Central Line tunnels is reduced, indicating that these tunnels, with cast-iron segmental linings, drain the surrounding ground. It should be noted that the profile for HP33 would also be expected to register slight under-drainage, but is close to hydrostatic because of suspected connectivity between the lower devices, as explained by Wan & Standing (2014b).

The total horizontal stresses measured by the spade cells (without correction of any factor of undrained shear strength) before the tunnel construction are given in Table 2 (note that the undrained shear strength determined from unconsolidated undrained triaxial tests on 100 mm dia. samples from boreholes in Hyde Park ranges from 150 kPa to 400 kPa at the level of the Crossrail tunnels (see Fig. 9 in Wan & Standing (2014a)). Using these uncorrected measured total horizontal stresses in conjunction with the known overburden stress and the measured pore water pressure at the depth of the spade cells, the at-rest lateral earth pressure coefficients (K₀) can be deduced and are also presented in Table 2. With the exception of the spade cell HP35, which measured unrealistically low total horizontal stress, the derived in situ K₀ values at the depths of spade cells HP36, HP37 and HP39 fall within a range from 1·9 to 2·4. Taking into account the likely over-read of the in situ stresses by spade cells, it is considered that these K₀ values are reasonable. Nearby, at Paddington, K₀ values of up to about 1·7 derived from self-boring pressuremeter measurements and independent suction measurements of undisturbed samples near the depth of the Crossrail tunnel axis level – that is, near the base of London Clay unit B (GCG, 2009) – corroborate the measured initial values of total horizontal stress from the spade cells. Typically, the design K₀ value of London Clay in central London, considering the historical erosion of top clays and subsequent reloading of superficial deposit, ranges from 1·0 to 1·5 (Burland et al., 1979).

Table 2. Steady-state pore water pressures and total horizontal stress (uncorrected by over-read factor) measured by combined spade cells, and estimated K0 before tunnel construction

For the multi-level VW piezometers and combined spade cells, during periods 2 and 4, hourly readings were taken and recorded automatically by data-loggers within each borehole headworks, except spade cells HP36 and HP37 where readings were taken manually using a hand-held VW readout unit three to four times a day owing to faulty data-loggers.

In the next sections the measurement results are presented in terms of either time or the distance, x_f, of the EPBM cutter-head relative to the monitoring line. The progression of both TBM1 and TBM2 with time is given by Wan et al. (2017a: Fig. 6).

Changes in pore water pressure, Δu, measured by the VW piezometer sensors in HP32 as TBM1 passed are presented in Fig. 5(a). In plan HP32 is about 10·8 m from the TBM1 centre-line, or one tunnel diameter from the extrados of the excavation. No measurement results are shown for the sensor at z = −24·5 m as it was faulty. Pore water pressures started to change at all elevations when x_f = −30 m. At about 25 m in front of the HP32 piezometers (x_f = −25 m), TBM1 stopped advancing for about 31 h because of problems with the spoil muck-away system. During this time the pore water pressures measured within HP32 over the depth of the TBM (−31 m < z < −42 m) fluctuated by a small degree: within about ±3 kPa. This incident did not cause significant settlement measured by the surface settlement points (Wan et al., 2017a: Fig. 8(a)) or subsurface rod extensometers (Wan et al., 2017b: Fig. 3).

Fig. 5. Change in pore water pressure measured in piezometer HP32 in response to westbound construction (period 2): (a) variation with TBM1 cutter-head distance; (b) depth profiles when the TBM1 cutter-head was approaching the instrument; (c) depth profile when TBM1 cutter-head was leaving the instrument

The pore water pressures at the measurement depths, except that furthest from the TBM at z = −12 m, increased as the TBM cutter-head approached, peaking when it was about 5 m beyond the instruments (x_f = 5 m). A maximum pore water pressure change of about +22 kPa was measured at the elevations near TBM1 crown and invert levels (z = −31·0 m and z = −37·2 m) which are equidistant from the axis of TBM1. The fact that these piezometers measured pore water pressure increases of a similar magnitude suggests that they are measuring an essentially undrained response to the TBM cutter-head approaching, despite the fact that the lower device is just within the more permeable A3ii sub-division of the LCF (see Fig. 2). Since the TBM face pressure (about 180 kPa on average) was lower than the overburden pressure at the tunnel axis level (about 650 kPa), it is believed that the pore water pressure increase in front of the TBM cutter-head was caused by ground arching in front of and around the TBM shield face (discussed in the section entitled ‘Mechanisms of EPBM tunnelling-induced pore water pressure changes and ground arching’).

As the TBM advanced and passed beyond the instruments, the pore water pressures at all elevations decreased as a result of the predominantly undrained unloading caused by the excavation until the TBM1 cutter-head was at a distance of about x_f = 25 m. The magnitude of the final change of the pore water pressure depends on the balance between the loading and unloading effects, the distance from the TBM, and also the permeability of the London Clay at the point of measurement. The undrained shearing of the overconsolidated clay would also generate additional negative excess pore water pressure. It appears that the shallowest sensor at z = −12 m was too far from TBM1 to be affected by the effect of the approaching cutter-head, but measured a slight decrease in pore water pressure due to the effect of ground unloading. The TBM1 tail leaving the instruments does not seem to have any significant effect on the pore water pressure (near x_f = 10 m) in HP32, although a very minor effect is noticed for the two sensors near the TBM crown and invert levels. By the time TBM1 is 60 m from HP32, all sensors have essentially stabilised, with the two shallowest ones being within ±5 kPa of their original values and the others with net negative changes (the largest being Δu = −16 kPa at z = −42·0 m).

Before stabilising, there are two points in time (at x_f distances of about +20 m and +40 m) when the sensors closer to the TBM appear to indicate marked, but small (<5 kPa), reductions in pore water pressure, the magnitudes of which decrease with increasing distance of the TBM from the sensors. The data points, which represent hourly readings, are closely clustered at these x_f distances and so coincide with times at which the TBM slowed down or was not advancing. The reason for the reductions is not known. Similar responses are also observed when the TBM slows down in some of the later plots.

Many of the observations described above are evident from vertical profiles of pore water pressure changes measured in HP32, which are plotted separately as TBM1 was approaching (Fig. 5(b)) and leaving HP32 (Fig. 5(c)). It is clear from these figures that the increase in pore water pressure in response to the TBM1 face approaching and then the subsequent reduction due to ground unloading was greatest at the TBM horizons.

Changes in pore water pressure measured by the VW piezometers in HP33 during period 2 are presented in terms of cutter-head distance in Fig. 6(a). HP33 is at a greater distance away from TBM1 than HP32 (about 16·0 m in plan from its centre-line) and therefore the overall response was of smaller magnitude. As with HP32, pore water pressures also started to change when x_f = −30 m. The effect of the TBM1 stopping at x_f = −25 m was not as obvious as measured in HP32 as HP33 was further from the tunnel. As the TBM1 cutter-head approached and passed beyond HP33, pore water pressure at the deeper four elevations gradually increased with a maximum Δu of about +15 kPa measured at z = −29·0 m near the TBM1 crown level. Only small changes (Δu < 10 kPa) were measured at the two shallowest sensor elevations (Δu ∼ 0, at z = −9 m). There was no discernible pore water pressure response above the TBM1 crown level to the ground unloading (i.e. pressure drop), as the shield tail passed HP33 (around x_f = 10 m), except a small reduction at z = −14 m. The response of the piezometer at this depth is slightly erratic at earlier TBM positions (e.g. x_f = −18 m) and so the output from it could be erroneous, especially given the piezometer responses above and below it (at z = −9 and −17 m). Therefore, overall the effect of stress relief or shearing due to excavation was not registered by the piezometers within HP33.

Fig. 6. Change in pore water pressure measured in piezometer HP33 in response to westbound construction (period 2): (a) variation with TBM1 cutter-head distance; (b) depth profiles when the TBM1 cutter-head was approaching the instrument; (c) depth profile when TBM1 cutter-head was leaving the instrument

Similar to the observation for HP32 (Fig. 5(a)), there were marked reductions in pore water pressure measured by the HP33 sensors at x_f =  + 20 m, when the TBM advance slowed down, but their magnitudes are even smaller than those measured in HP32.

Profiles of Δu with depth when TBM1 was approaching and passing beyond HP33 are shown in Figs 6(b) and 6(c), respectively, and reflect many of the observations made above (especially regarding the piezometer at z = −14 m: if the data from this device were omitted the profiles would be far more uniform). As discussed in Wan & Standing (2014b), the lowest three piezometers within HP33 were found during the pre-construction monitoring to be interconnected after installation. Despite this, independent immediate pore water pressure changes still seem to have been measured reliably by these three lowest sensors in response to the rapid loading and unloading during tunnel construction, with the pressure increase decreasing with distance from the tunnel crown.

Multilevel piezometer borehole HP34 is 5 m in plan from the extrados of the westbound Central Line tunnel and was located directly above TBM1 when it passed, with the deepest sensor at z = −29·0 m, just 2 m above its crown. Consequently the measured pore water pressures might have been expected to be more pronounced as TBM1 approached and passed.

It can be seen from the measured values of Δu presented in Fig. 7(a) that this was not so (results from the piezometer at z = −26 m are not shown as it was faulty). Increases in pore water pressure were less than 5 kPa as TBM1 approached HP34, compared with an increase of up to 22 kPa and 11 kPa measured at HP32 and HP33, respectively, which were more than 10 m horizontally from the TBM1 centre-line. The reduced increase in Δu probably occurs because of the close proximity of the adjacent existing Central Line tunnels, which would have affected the ground loading/unloading pattern and the development of any ground arching in front of the advancing TBM cutter-head.

Fig. 7. Change in pore water pressure measured in piezometer HP34 in response to westbound construction (period 2): (a) variation with TBM1 cutter-head distance; (b) depth profiles when the TBM1 cutter-head was approaching the instrument; (c) depth profile when TBM1 cutter-head was leaving the instrument

Pore water pressures started to decrease when the cutter-head was about 7 m, or about one tunnel diameter, in front of HP34 (x_f = −7 m). The greatest recorded change was at z = −29 m, above which values decreased with increasing elevation above the TBM1 crown. Pressures continued to decrease until about x_f = 4 m as negative excess pore water pressures were generated as a result of unloading around the front section of the shield body (i.e. soil closing into the annular void created by the tapering shield) and possibly also shearing of the overconsolidated clay. Subsequently, at elevations closer to the TBM1 crown (i.e. z = −22 m and −29 m), pore water pressures rebounded by up to 35 kPa. This could be a consequence of the tail grout pressure compressing the ground (even though the main effect of the grouting would be expected directly behind the rear of the shield) and correlates exactly with what was observed from the response of the extensometers directly above TBM1 as reported by Wan et al. (2017b: Fig. 3(b)). (Note that the comparison made relates to extensometer HP20, which is about 38 m from HP34. However, the response of extensometer HP26, directly adjacent to HP34, was very similar but with a less marked rebound (this response was not presented in the Wan et al. (2017b) paper but was in the thesis by Wan (2014)).) Ground heave from tail grouting occurred at about x_f = 7 m, in advance of the back of the shield passing beneath the measuring point, followed by a rapid settlement. Another explanation for this increase in Δu (rebound) before the back of the shield passes beneath HP34 can again be provided by a mechanism of ground arching. Ground arches would form around locations where ground losses are greatest and the ground has softened or ‘yielded’ (i.e. at the shield tail where tail void closure starts to occur before the grouting comes into effect). The ground arching would compress the ground in front of and behind the shield tail, leading to increases in pore water pressure (discussed further in the section entitled ‘Mechanisms of EPBM tunnelling-induced pore water pressure changes and ground arching’).

Similarly, in terms of pore water pressure changes, following the rebound, Δu values dropped rapidly at the deeper elevations as the rear of TBM1 passed beyond HP34 (Fig. 7(a) from x_f = 7 m onwards). This reduction in Δu occurred as the shield was advanced and the ground lost the support of the shield body, leading to further ground loss into the gap between the tail-skin and the extrados of the newly erected tunnel linings (tail void closure). A maximum reduction of pore water pressure of about −75 kPa was measured at z = −29 m when x_f = 13 m. A similar response in terms of displacement was observed (Wan et al., 2017b: Fig. 3(b)). It is evident that, for the ground near the shield tail (x_f = 10 m) the effect of tail void closure (stress relief) significantly outweighs the effect of tail grout pressure (compressing the ground). Soon afterwards the pore water pressures began to recover as the newly erected tunnel linings, in conjunction with tail grout setting and becoming stiffer, started to support the annulus between their extrados and the ground and also because of the development of ground arching across the rear of the shield. For the three operational sensors above the TBM1 crown (−29 m < z < −17 m), the final pore water pressures were smaller than those prior to the construction.

Profiles of pore water pressure changes with depth for the periods when TBM1 approached and passed beyond HP34 are shown in Figs 7(b) and 7(c).

As can be seen from Fig. 1, the spade cells are all very close to the eastbound tunnel alignment (therefore at some distance away from TBM1). Changes in pore water pressures, Δu, and total horizontal stresses, Δσ_h, measured by the spade cells in response to the westbound construction are shown in Fig. 8. Spade cell HP39 was located at the same y–z position as the deepest piezometer of HP33: its pore water pressure response, as shown in Fig. 8(a), has a similar pattern, with an increase in pressure (cf. Fig. 6(a)), but of slightly greater magnitude than the deepest piezometer in HP33 as TBM1 progressed. A potential reason for the differences in magnitude is that the piezometer within the spade cell responds more simultaneously with total stress changes as it is in direct contact with the ground (from pushing in the device) compared with the multi-level piezometer installed within a grout column. The other three spade cells HP35, HP36 and HP37, located at the elevation of the TBM axis, also exhibited similar pore water pressure responses but of smaller magnitude. Spade cell HP35, closest to TBM1 (11 m from its centre-line or about one diameter from the tunnel extrados), measured a slight drop of pore water pressure at x_f ≈ 20 m, compared with the other three devices which showed a small increase.

Fig. 8. Change in (a) pore water pressure and (b) total horizontal stress measured in spade cells HP35, HP36, HP37 and HP39 in response to westbound construction (period 2)

Total horizontal stress changes, Δσ_h, transverse to the axis of the tunnel are shown in Fig. 8(b). Those measured by HP35 were the largest in magnitude, it being the closest spade cell to TBM1 (although still 11 m from its axis). At this location, the total stress initially increased by 20 kPa in response to the approaching cutter-head (probably associated with ground arching in front of the cutter-head). The unloading effect in the vicinity of the rear of the shield as it passed is evident as a rapid, large drop in total stress (for x_f > 10 m). A maximum decrease of about −80 kPa was measured at x_f = 20 m, before σ_h started recovering. The net total stress reduction was measured to be about −50 kPa, indicating the overall effect of the tunnel excavation and the newly erected tunnel lining supporting the ground. Additionally, as HP35 is located within the very silty B1 sub-division, some pore water pressure dissipation might occur in the short term.

At the other spade cells (HP36, HP37 and HP39) further away from TBM1, initial increases in total stresses were smaller (up to +10 kPa) as TBM1 advanced and passed, followed by negligible subsequent drops and rises after the initial increases (less than ±5 kPa) compared with the response observed for HP35.

HP32 lies very close to the tunnel construction, being about 1·5 m (less than half of a tunnel radius) horizontally from the TBM2 extrados, resulting in a much greater pore water pressure response than that from the westbound tunnel construction. Changes in Δu measured by the five operational piezometers in HP32 as TBM2 passed are shown in Fig. 9(a), where it can be seen that Δu increased at all elevations, except the shallowest one, when TBM2 was at x_f = −40 m, rising rapidly by x_f = −15 m. Peak increases were measured within −10 m < x_f < −3 m, depending on the individual piezometer elevations. A maximum change of about +50 kPa was measured at elevations near the TBM2 crown and invert levels (z = −31·0 m and −37·2 m) in response to the approaching TBM2 cutter-head and the development of ground arching. Pore water pressures then dropped rapidly by up to 130 kPa near x_f = 0 m, from the unloading and shearing associated with the excavation and advance of the shield. These Δu values suggest that conditions were essentially undrained. The effect of tail grouting (average pressure of about 80 kPa) and the development of ground arching is evident from pore water pressures starting to recover between 5 m < x_f< 10 m with rebounds ranging from 30 kPa to 50 kPa at elevations close to TBM2 (as noted for TBM1, these increases occur well in advance of the tail skin reaching the monitoring position). This correlates very well with the ground being temporarily pushed outwards by the tail grout in advance of the TBM2 tail skin as measured by inclinometer HP9, which was at a similar offset distance to the TBM2 axis but on the other side (Wan et al., 2017b: Figs 12 and 13).

Fig. 9. Change in pore water pressure measured in piezometer HP32 in response to eastbound construction (period 4): (a) variation with TBM2 cutter-head distance; (b) depth profiles when the TBM2 cutter-head was approaching the instrument; (c) depth profile when TBM2 cutter-head was leaving the instrument

As the shield tail passed beyond HP32, x_f> 10 m, there was a second rapid reduction as a result of further unloading (tail void closure). The maximum net reduction of pore water pressure was about −150 kPa measured below the tunnel invert at z = −42 m when x_f≈ 15 m. Pore water pressures then started to recover by different degrees from the tail grout setting and becoming stiffer as TBM2 advanced further.

The depth profiles of the change in pore water pressures measured in HP32 as TBM2 was approaching and leaving are shown in Figs 9(b) and 9(c), respectively, and reflect the trends described above. Being nearest to the TBM location, the sensors at z = −31 m and −37·2 m measured the greatest pore water pressure increase when the TBM cutter-head was approaching (when x_f = −5·4 m). As TBM2 passed beyond HP32, the piezometer at z = −42 m, being the deepest one, experienced the largest unloading as the ground above it was excavated and hence measured the greatest pore water pressure reduction.

The Δu values measured by the piezometers in HP33 as TBM2 passed beneath them are presented in Fig. 10(a). The measured responses to the eastbound construction show a very similar pattern to those of HP34 in response to the westbound construction (see Fig. 7(a)), but are generally of larger magnitude. In particular, the pressure increase measured by HP33 ahead of the cutter-head was more pronounced, probably because ground arching in front of the cutter-head could develop freely without any interference from the existing Central Line tunnels. Pore water pressures increased (up to +20 kPa) as the TBM2 cutter-head approached, followed by: a rapid drop of pore water pressure near x_f = 0 m; subsequent rebound between 5 m < x_f < 10 m; a further drop between 10 m < x_f< 15 m; and subsequent partial recovery. The greatest net reduction of pore water pressure was about −190 kPa at z = −29 m when x_f = 15 m. Profiles of Δu with depth as TBM2 was approaching and leaving are presented in Figs 10(b) and 10(c), respectively. The deepest piezometer at z = −29 m, about 2 m above the TBM crown, recorded the greatest Δu. As observed for the westbound tunnel construction, although the lowest three piezometers in HP33 were interconnected, it seems that the immediate pore water pressure responses to eastbound tunnel construction were still measured independently.

Fig. 10. Change in pore water pressure measured in piezometer HP33 in response to eastbound construction (period 4): (a) variation with TBM2 cutter-head distance; (b) depth profiles when the TBM2 cutter-head was approaching the instrument; (c) depth profile when TBM2 cutter-head was leaving the instrument

Pore water pressure changes measured within HP34 in response to eastbound tunnel construction, as shown in Figs 11(a)–11(c), were much smaller than those of HP32 and HP33, it being further away from TBM2 (18·6 m horizontally from its centre-line). Only the deepest sensor at z = −29 m measured a gradual increase of pore water pressure up to +8 kPa during the passage of TBM2. Negligible changes were registered by the piezometers at other depths. The magnitude of the Δu response is even smaller than that measured within HP33 in response to the westbound tunnel construction (cf. Fig. 6). A likely reason for the smaller response at HP34 is the combined presence of the existing Central Line tunnel (see Fig. 1) and, perhaps more predominantly, the newly constructed Crossrail westbound tunnel. Both are stiffer than the ground and hence ‘attract’ more total stress changes induced by the eastbound tunnel construction. As a result the ground at HP34 would have experienced a smaller total stress change (and hence smaller excess pore water pressure) than the ‘greenfield’ ground at HP33 when the westbound tunnel was being constructed.

Fig. 11. Change in pore water pressure measured in piezometer HP34 in response to eastbound construction (period 4): (a) variation with TBM2 cutter-head distance; (b) depth profiles when the TBM2 cutter-head was approaching the instrument; (c) depth profile when TBM2 cutter-head was leaving the instrument

Data from the measurements of Δu and Δσ_h from the spade cells in response to the eastbound tunnel construction are presented in Fig. 12. The data-loggers serving HP36 and HP37 became faulty during the tunnel construction period and so measurements were taken manually three to four times a day using a hand-held VW readout device and thus some critical readings might have been missed (compared with the hourly data logging). As shown in Fig. 12(a), values of Δu measured about 1·5 m either side of the eastbound tunnel extrados (HP35 and HP36) and about 2 m above the eastbound tunnel crown (HP39) showed almost the same pattern, with very slight increases (up to +5 kPa) as the cutter-head approached them (average face pressure = 200 kPa), followed by a consistent reduction of pore water pressure as TBM2 passed. The magnitudes of change were much larger than for TBM1, as with the TBM2 drive the relative distances were much closer, but the sense of change was completely opposite with decreasing pore water pressures observed compared with the increases seen when TBM1 passed. This suggests that the effects of unloading and shearing are much greater in the close vicinity of the TBM (TBM2 in this case, around which the instruments were closely clustered). Additionally, the rebound of pore water pressure, induced by the tail grouting as observed within the piezometers in HP32 and HP33 at corresponding y–z positions at x_f = 10 m was not observed. In this case the pore water pressure responses measured at the spade cells do not seem as sensitive as those measured by the multi-level piezometers (HP32 to HP34). The reason for this is not known but might be associated with the contact condition (e.g. smeared or voided interface) between the soil and the piezometer filter on the spade surface. The final reduction in pore water pressure at these spade cell locations was about −100 kPa. Values of Δu measured by spade cell HP37 further away from TBM2 show responses of much smaller magnitude, being less than ±5 kPa and so almost negligible. It is worth noting that, despite the difference in sensitivity, the magnitudes of the final change of pore water pressure after TBM2 has passed (x_f = 60 m) measured by the deepest piezometer of HP33 and the spade cell HP39 (at the same relative position to TBM2) were comparable to each other, being about 100–115 kPa.

Fig. 12. Change in (a) pore water pressure and (b) total horizontal stress measured in spade cells HP35, HP36, HP37 and HP39 in response to eastbound construction (period 4)

There is a marked Δu reduction of up to 20 kPa measured in HP35, HP36 and HP39 at about x_f = 40 m, which again is associated with a period when the TBM advance slowed down (clustering of hourly readings). However, at the same time the corresponding Δσ_h values (transverse to the tunnel axis) measured by these spade cells were much less pronounced. This might imply that the marked changes in pore water pressure at x_f = 40 m originated mainly from the total horizontal stress change in the direction parallel to the tunnel axis.

An increase of total horizontal stress of up to +80 kPa was measured in HP35 prior to the cutter-head reaching the spade cells, probably as a result of the development of ground arching ahead of the approaching cutter-head (Fig. 12(b)). Very small total stress increases can be observed from the manual measurements at the other spade cell HP36, at the same offset but on the other (northern) side of the tunnel extrados, but no manual measurements were made at the time when the peak was observed in HP35. It therefore seems likely that greater Δσ_h values would have been observed at HP36, and perhaps HP37 (but judging from the trends, not to the same degree as HP35). However, no total horizontal stress increase was measured in the spade cell HP39 (hourly data-logged) directly above TBM2 crown, as might have been anticipated.

All four spade cells recorded a reduction in total horizontal stress as the TBM2 cutter-head reached them (x_f = 0 m), as a consequence of ground loss and stress relief due to the excavation. Readings from both HP35 and HP39 indicate a rebound of total stress between 5 m < x_f< 10 m from tail grouting and the development of ground arching, reaching a peak, followed by a further drop between 10 m < x_f< 15 m as the ground closed in towards the newly erected lining. It appears from the data available for HP36, and perhaps HP37, that the same trends would have been observed but have been missed because of the much lower frequency of manual measurements. Total stresses recovered partly as TBM2 advanced further from 15 m < x_f< 20 m as a result of the tail grout setting and gaining stiffness, after which they stabilised.