A TBM entering a shaft eye frequently passes through a mechanical seal to resist water ingress between the machine and shaft. In granular ground below the water table, a breached seal may cause a catastrophic inundation and flooded shaft.
Inundations occur when the ground is mobile. Over-saturated granular ground flows through the breach, and pumping to reduce water levels in the shaft only induces further ground inflows. This ground movement occurs in granular soils from the finest sands to large gravels – whenever the hydrostatic head of free water and grain size is appropriate, granular ground becomes unstable and mobile. The use of polyurethane resin grouts can often stabilise these disastrous situations, enabling early pumping of the shaft and recovery of the TBM. This paper presents two successful cases of flooded TBM recovery: a reception shaft situated between a main road and dockside railway, and a launch shaft in marshy ground adjacent to a major river.
Financial aspects
The financial consequences of a breach are significantly different if the TBM is entering or exiting a shaft. If entering, the machine is essentially isolated from water ingress as the tail skin and tunnel build behind the TBM effectively cut off any flow paths to the flooded shaft. In this situation, the TBM internals are protected and containing the breach allows the shaft to be pumped out and the machine re-started and complete its drive.
When launching from a shaft, the electrical and hydraulic circuits within the TBM are exposed and unprotected from water incursion if the shaft floods. This can cause much damage and, before the machine can be re-started, a costly strip down and rebuild of these circuits is necessary.
Any unscheduled stop of a TBM is an expensive disruption to programme, with further difficulties arising if the ground settles. Inundation into a shaft can cause localised depressions at ground level and, if near to property or part of the infrastructure, structural damage may result. This consequential loss only increases the overall financial burden. It is therefore prudent to seek a means to recover such situations with the minimum downtime. The adept use of resin injection can achieve a speedy result, limiting the loss of production time. Resin injection also provides an extremely cost-effective method of controlling a breached seal.
Technical considerations
Using resins in such situations requires an experienced, competent injection engineer with a trained crew; additionally, the client must be receptive to innovative grouting techniques. There is no universal remedy to recovery operations. Each job is individual. A detailed inspection of a flooded shaft is almost impossible, yet the injection engineer must quickly grasp an understanding of site conditions if the operation is to succeed. Events require separate assessment and techniques applied accordingly, as the following cases demonstrate.
Project 1 – TBM entering shaft
The first project relates to a large diameter Lovat driving through marine gravel into the soft eye of a reception shaft fitted with a proprietary seal. The seal was inflated with cement grout to provide intimate contact with the emerging TBM. As the TBM entered the shaft, the seal ruptured, the most likely cause being broken segment arisings failing to clear the cutter face. The TBM advance was immediately stopped, but heavy inflows through the breach washed in quantities of gravel and quickly flooded the 25m deep shaft to within 2m of the top.
Attempted pumping out only caused more ground to enter the shaft, an undesirable situation with the shaft neatly sunk between a railway and main road.
As the shaft water levels could not be lowered, it was necessary to employ divers to investigate the breach and report to the injection engineer at surface level. The divers, operating in zero visibility, were hampered by equipment at the shaft bottom and washed in ground. Therefore the injection engineer’s interpretation of the diver’s report needed to be tempered by empirical judgement and experience; in these conditions there is often no possibility of calculating void sizes and grout take up.
On this project, the breach occurred to the lower right side of the seal, a position relating to approximately 4 o’clock when facing the cutter head. The diver reported being able to push a steel probe horizontally through the breach for over 1.5m before encountering loose, unstable gravel. It was not possible to establish the vertical extent of the void behind the seal; the diver could only report it exceeded the horizontal dimension.
The injection engineer determined that the primary task was to fill the void behind the seal and contain further ground loss. The importance of fully controlling water flows was secondary, as inflows may occur elsewhere through the seal.
The injection engineer required a resin grout product that would not be diluted or destabilised by water, would act as an efficient void filler and provide structural rigidity to prevent further ground loss when the shaft was pumped out. The resin also needed to be fast reacting, to ensure it was not dissipated after injection. Such properties are available from two component polyurethanes and suitable products were sourced virtually overnight through Stabila UK Ltd.
Two component polyurethanes consist of a polyol and isocyanate that start to react on mixing. The selected resin reacts from a liquid to a stable closed-cell foam, which exhibits excellent structural properties and resistance to water movement. The reaction is very rapid and it is necessary to pump the two components through individual lines to the point of injection where the lines merge into a static mixing chamber. Thereafter, a single line leads to an injection lance from which the reacting resin emerges.
As the shaft was flooded, the resin pump needed to remain at ground level. The two resin lines, of sufficient length to reach the breach, the mixing chamber and lance had to be charged with resin before lowering this unwieldy bundle to the diver.
The injection engineer directed the diver to insert the injection lance at the lowest possible point in the breach and then commenced resin injection. Reacting resin produces foam with a lower specific gravity than water; the foam therefore rose to the upper limits of the void whilst injection continued at the lowest point. As the foam expanded, it also penetrated the gravel and on final cure provided a structural infill to the void and stabilising medium to the loose gravel face.
The process continued on a stop-start basis, with the injection engineer monitoring progress and allowing sufficient resin reaction time to minimise undue resin loss into the gravel or open face of the breach. Finally, on detecting indications that the process was accomplished, the injection engineer stopped injecting.
On retrieving the lines and diver, the shaft was pumped out with the resin containing all further ground inundation. Some heavy residual water inflows were controlled by further injection of polyurethane resins and the TBM was then successfully driven into the shaft. The whole operation was completed in just five days, including initial site inspection to de-mobilising.
Project 2 – TBM exiting shaft
In this instance, an Iseki Unclemole heading a pipe jack was launched from a shaft through a “top hat” into fine estuarine, granular soils with the ground water level almost at the surface. The “top hat” incorporated a flexible seal, but a breach occurred only a few metres into the drive. The resulting inundation flooded the shaft in a matter of minutes, simultaneously infiltrating the internal components of the TBM.
Site hearsay claimed some holding down bolts failed, allowing the “top hat” to move away from the shaft. Oxford Hydrotechnics proposed a scheme of using divers to investigate conditions within the shaft before preparing a stabilising package based on resin injection.
However, the client was swayed by a suggestion of de-watering to gain access to the shaft, even though the site was close to an estuary with large sea-going vessels regularly passing. Suffice to say, de-watering proved ineffective and, after several weeks of abortive pumping, the client instructed Oxford Hydrotechnics to mobilise and investigate conditions within the shaft.
Working in a very congested environment with zero visibility, divers examined the breach, using information proffered by the client’s site personnel of where the seal failed, and located a gap between the shaft and one side of the “top hat”, although the holding bolts appeared intact. Ground from the inundation, and various pieces of equipment, prevented access to the lower sections of the “top hat”.
Being unable to equate these findings to the rate of inflows as depicted by the client, or to the volume of ground discovered in the shaft, the injection engineer instigated a systematic sequence of investigations that required the diver to enter the pipe and drill a series of holes through the pipe between the TBM and shaft.
By probing with a steel rod, a substantial void was discovered under the pipe, adjacent to the shaft. Further investigation revealed the void extended up the pipe sides in a crescent-like profile (Figure 1). Subsequent probe holes revealed the void extended back from the shaft, with reducing dimen-sions towards the TBM.
This project differs from the job described previously in a number of aspects, and the injection engineer determined his task was to both fill the void and control the water movement, as the very small grain size of the ground made it susceptible to flow. In gravels, larger material can interlock and pack tight to halt ground loss through a restricted aperture; with small grain sizes this does not happen, the mobile ground virtually flows without constraint.
The resin was required to withstand direct injection into water, be a fast reacting void-filler but not a structural material. In this situation filling the void around the pipe with a structural resin would risk resin and ground adhering to the pipe, effectively stopping any further movement to the early pipe jacking process. Also, the small grain and inter-grain void sizes would restrict penetration by more viscous structural resins. A single-component, hydrophilic polyurethane was selected as the main stabilising product; this resin readily reacts with water to produce a gel within approximately 20 seconds. A multi-component injection pump was necessary to pump two components (resin and water) through individual lines to a static mixer near the point of injection. The length of single line from the mixer to the injection point was determined by the requirement to have reacting resin emerging from the line.
With the resin pump at surface level, the individual lines, mixing chamber and single line, were purged and charged before lowering to the diver in the shaft. However, suspecting other problems, the injection engineer held other polyurethane resins in reserve.
The void capacity under the pipe was calculated from information obtained by the diver’s probes and the requisite volume of resin/water mix injected through packers set in the previously drilled probe holes. On partially pumping out the shaft, heavy water inflows remained through the gap between “top hat” and shaft. A grout bag manufactured on site was installed by the divers and injected with one of the reserve resins, a hydrophobic polyurethane, to control this inflow.
The shaft was then pumped out, but after 5-10 minutes a slight flow of murky water was observed through the sludge at invert level, below the “top hat”. The flow quickly developed into an inundation that once more flooded the shaft. Divers again investigated within the the “top hat” but failed to establish any change in conditions that would indicate why the incursion occurred, again the divers could not examine the lower sections of the “top hat”.
After re-considering the findings from the initial examination, a further sequence of resin injection was devised using both hydrophobic and hydrophilic polyurethane resins.
The hydrophobic resin produces a tough, flexible closed cell foam and was injected by means of ports drilled through the shaft around the “top hat” to reduce any water flows immediately adjacent to the shaft. Injection was limited to the upper section of “top hat” as various impedimenta prevented access to the lower section. A blend of accelerated hydrophobic and hydrophilic polyurethanes was made on site and injected into the void previously detected under the pipe; this combination of resins reacted to stiffen the gel already injected.
The shaft was again pumped out, but water still flowed in at invert level. However, the water flow was now clear and without any evidence of ground incursion and no further inundation occurred. This inflow was reduced by injecting “Hydrablok” polyurethane resin, a system which reacts instantaneously, to produce spaghetti-like strands of cured resin that quickly builds, to choke off free-flowing water. Any residual inflows were now easily pumped by a small submersible.
Several cubic meters of washed in ground were cleared from the shaft before it was possible to inspect the “top hat” at invert level, where an examination revealed a section of segment between shaft base and pipe was missing (Figure 2). It is surmised that the segment was damaged during the launch and then totally dislodged as the pipe jack progressed. The gap was sufficiently large to permit the high rates of inflow and, being at invert level, was initially beyond the diver’s examination.
This project was spread over two phases: initially two days for divers to investigate under the direction of an injection engineer; the subsequent injection operation, including the second incursion, taking just five days.
Conclusions
These two projects clearly illustrate the versatility of properly engineered polyurethane injection, but there is no simple formula for recovery operations. Each job needs individual assessment and the interpretation of any diver’s report must be tempered by empirical judgement and experience.
Soils investigation reports may not be of any great benefit, as the fines are washed out during inundation, thus locally changing the ground; the SI information that may possibly assist in calculating grout take-up is therefore considered unreliable.
The second case shows how the injection engineer may need to revise procedures and change materials as works evolve. Any method statement must be adaptable, to enable these changes to take place.
Polyurethane resins offer unique properties in controlling ground water and stabilising mobile ground. These properties can be strategically applied in the recovery of inundations, flooded shafts and similar disastrous situations below ground. In such situations, clients need to be receptive to lateral thinking and permit the injection engineer to ply his trade.
Employing an experienced injection engineer and specialist crew who utilise innovative techniques are essential if the full benefits of these versatile resins are to be gained.
Related Files
Fig 1 – Section through pipe approximately 300mm from shaft – from initial investigations
Fig 2 – Schematic section through shaft eye
