The $110M North Downs tunnel, Contract 410, is one of the six major civil engineering contracts on the UK’s first high-speed railway, CTRL Section 1, from the Channel Tunnel north through Kent to Fawkham Junction. The 3.2km long single bore, twin-track tunnel passes under the North Downs between the Medway Towns and Maidstone at depths of up to 100m. The tunnel passes deep under Bluebell Hill village and crosses under the A229 dual carriageway three times before emerging at the country portal to the north of the M20.
Rail Link Engineering (RLE) carried out the design and management of the CTRL. The contract was awarded to Eurolink in October 1998. Construction started in April 1999, with completion in July 2001, five months ahead of programme. Cost savings of around $14M were made through design refinement.
Tender design
The tunnel cross-section was mainly determined by the need to ensure aerodynamic pressures generated by trains passing at 300km/h would be held to an acceptable level. Through extensive computer modelling the required free internal cross-sectional area was defined as 96.2m2. This was obtained with an internal height of 10.7m and width of 13m. The cross-section also had to include two platform-level evacuation walkways and a central barrier, to keep a derailed train from crossing on to the other track. These requirements, together with the space needed for the primary and secondary linings, gave a maximum total excavation area of over 165m2.
The tunnel was constructed entirely through chalk. From the London portal, it passes through progressively older strata – from Lewes Chalk to the New Pit Chalk, then Holywell Chalk, with the Lower Chalk entering the invert close to the country portal. As the tunnel moves through this sequence the chalk increases in unconfined compressive strength from around 2MPa to 4.5MPa. It also increases in density and becomes more massive.
The tunnel design was developed by RLE prior to tendering. The secondary lining consisted of reinforced concrete throughout, with invert up to 1,700mm deep and a 500mm-thick vault. The reinforcement was generally required to provide structural capacity over much of the tunnel length, and minimum reinforcement in other areas to control cracking.
Development of the design after tender was driven by both a formal value engineering process and informal design development. A value engineering workshop was held soon after the tender was awarded. A brainstorming session generated 68 ideas which were developed and combined into options for further consideration. Each option was examined to determine the cost and programme saving. The main proposals were:
It is worth noting that the value engineering exercise underestimated the savings that were eventually made when the ideas were taken forward.
Further design refinement continued throughout tunnel construction and the presence of the design team on site meant that new ideas could be evaluated and incorporated, including: continuous pouring of the reinforced invert slab; the use of longer shutters for the secondary lining vault concrete pours; and optimising the concrete strength required for striking the vault shutter to improve cycle times.
The refinement of several geotechnical parameters became a critical activity and a contract for additional testing was awarded for three deep boreholes along the line of the tunnel, and laboratory testing. Two key parameters singled out for further investigation, were the in situ stress ratio and creep properties of the chalk.
The estimation of the in-situ stress state for the tender design was based on the results of pressuremeter tests in boreholes along the route and a review of the published information for the Channel Tunnel. Based on assessment of this information, K0 values in the range 0.5 to 1.5 were selected for the purposes of design, on the assumption that lower values of up to 1.0 existed at the tunnel horizon in areas of shallow cover because of the effects of weathering. Higher values of 1.5 were assumed for the tighter chalk to be tunnelled through under the main escarpment at Bluebell Hill.
The design team had been aware of the impact of this range of K0 values on the design. However, the high cost and risk associated with using a more refined technique – hydrofracture testing – meant this option was not considered appropriate during the RLE design phase prior to tender award.
The use of hydrofracture testing was a fundamental part of the value engineering proposals, to clarify the in-situ stress state and enable the flat invert profile to be adopted. Hydrofracture testing involves injecting fluid, in this case water, into sealed-off borehole intervals to induce and propagate a hydraulic fracture in the rock mass. The pressure in the fluid is recorded so that the stress at the time of fracture can be calculated. The orientation of the fracture can be determined using impression packers lowered down the borehole after completion of the test. The specialist testing was undertaken using a high-pressure wireline packer testing system.
The tests indicated the mean minimum principal horizontal stress ratio was 0.75 and aligned transverse to the tunnel alignment. Based on this information K0 values in the range 0.5 to 1.0 were adopted for design of the secondary lining.
A series of long-term laboratory creep tests were carried out on samples recovered from the deep boreholes. However, the results of the creep testing broadly confirmed RLE’s original design assumptions and consequently no reduction in long-term creep loading was possible.
Primary lining design and review
The contract documents required the contractor to provide a fully designed primary lining, in accordance with BS8110, for a series of defined lengths, or zones, of tunnel. The design was to be adequate for the worst loading conditions anticipated within the zone. Support requirements for a zone could not be reduced without further design work, but additional measures could be added in areas of unexpectedly adverse geology.
Although this arrangement split the design responsibilities it was considered important to allow the contractor to design the primary lining as it influenced so much of its work. The problems of split responsibilities were overcome by placing a small RLE design team on site, responsible for a review of the primary lining as well as design of the secondary lining.
The primary lining design was carried out by Beton-und-Monierbau in Innsbruck and checked by Miller Civil Engineering in Rugby. The tunnel excavation was divided into three sections to give an excavation sequence of top heading, bench and invert. The support is mainly provided by the shotcrete lining, with rock dowels and spiles used to reinforce the ground. Lattice arch girders were used to provide profile control and reinforcing mesh to improve constructability. The design provided adequate lining support for the most onerous loading conditions, and also contained an array of options for providing suitable face support.
A full review of the design was conducted on site by RLE, predominantly using FLAC, a finite difference computer analysis program. The tunnel excavation and the installation of the primary lining in the heading, bench and invert was modelled in sequence, allowing loads in the lining to be determined. A hypothetical modulus approach was adopted for the primary lining to represent the plastic yielding of green shotcrete, with a routine written in FLAC that modelled plastic behaviour when a stress level of 5MPa was reached. Empirical assessment of the design was also carried out using the CSIR support classification system and Hoek-Brown support reaction curves.
Tunnel excavation and primary lining
Tunnel excavation commenced at the country portal end of the tunnel on 16 April 1999. Initially, only short lengths (up to 60m) of the top heading, bench and invert were excavated to form the first 100m of tunnel lining into a closed ring and provide stability in the portal areas. Excavation then concentrated on progressing the top heading. A similar sequence was followed at the London portal, but the start of excavation was delayed as significantly more stabilisation works were required on the portal slopes because of the presence of solution features in the chalk. The early ring closure was completed for the first 230m to provide stability for a section with low cover and poor ground conditions under the Buckmore Park karting track.
Tunnel excavation was initially carried out using a Liebherr 932 excavator and as the strength of the chalk improved, Paurat roadheaders were used to excavate the face. Shotcrete was placed using a Normet Spraymec robot, and rockbolts and spiles installed using an Atlas Copco boomer.
Once the learning curve had been overcome, six advances of between 1-2m were completed every 24 hours in the top heading. Each advance cycle consisted of excavation and muck away, then spraying the lining, and finally bolting and spiling. The primary lining was built up in three layers, with an initial layer (or flashcoat) of at least 50mm. This was covered with a bulk layer of a least 100mm after a sheet of mesh had been fixed in place.
When required, a second layer of mesh was placed before the final layer, which covered the ends of bolts to provide a uniform shotcrete surface. The bench and invert drives both used spilt faces to optimise advance rates. When the invert construction began, the redesign of the secondary lining was complete, allowing the flat profile to be constructed.
The following advance rates were obtained:
These rates were significantly faster than those assumed in the tender and led to the commencement of the secondary lining invert at the country portal four months ahead of the original programme. The potential incorporation of the primary lining into the permanent works meant that additional measures were taken to ensure the shotcrete was of sufficient quality and durability. This involved strength testing of the shotcrete at 56 days, in addition to the standard three, seven and 28-day tests, as well as permeability testing.
Monitoring review and back-analysis
A monitoring regime provided information on the movements of the tunnel lining and ground around and above the tunnel. Instrumentation included surface settlement points, in-tunnel and surface extensometers, and tunnel deformation monitoring.
The surface settlement was monitored during construction, with a total of 179 settlement points that were measured using precise levelling techniques. Settlement was measured on eight transverse arrays, mainly situated in the shallow cover areas, above the tunnel centreline, and at additional points in Buckmore Park. The transverse settlement measured during construction gave good approximation to a typical settlement trough following a Gaussian distribution where the cover to the tunnel is less than 30m.
Surface extensometers were installed above the tunnel centreline in the three deep hydrofracture boreholes and four further holes in shallow cover sections near the portal. The extensometers, from Geotechnical Instruments, consisted of a series of pneumatically released spider magnets installed into previously drilled boreholes. The top of the borehole was also regularly surveyed to determine movement of the ground level, so that the movement of each magnet could be calculated.
Five arrays of rod extensometers were installed from within the tunnel. Each array consisted of three arms, one extending vertically up from the crown and the others at the base of the top heading, radial to the tunnel. Within each arm were 3m, 6m and 9m long extensometers and, in the three arrays in high cover sections, a further 15m long extensometer. The extension or shortening of each extensometer along its length could be read to an accuracy of at least 0.1mm.
The time taken to install the extensometers meant most movement in the top heading was missed, but in deep sections there was good agreement with tunnel deformation monitoring for bench and invert construction. In shallow cover sections there was little differential along the length of the extensometers.
Monitoring of the in-tunnel deformation was carried out using precise 3D surveys of arrays of Bioflex targets at intervals of between 5m and 40m. Readings accuracy, governed by quality of targets and atmospheric conditions, was ±0.5mm. The deformation monitoring results were post-processed using the Dedalos tunnel deformation program, allowing the presentation of the vertical, transverse and longitudinal tunnel lining movements at each array, against time.
The tunnel deformation monitoring, taken together with the surface monitoring points above the tunnel centreline, showed a marked reduction in movement once the depth to tunnel axis exceeded 1.8 tunnel diameters in undisturbed ground. In areas near the country portal, where there was up to 15m of made ground, the change occurred at a depth to axis of 2.2 tunnel diameters. With information from the other instrumentation, it was considered that the behaviour of the ground changes from a “chimney” type loading (typical in areas of shallow cover) to that of a rock arch above the tunnel (typical deep cover behaviour).
Monitoring results, along with geological logs of the tunnel face, were discussed daily, allowing adjustments to the advance length and face support details. Results were also used to confirm the secondary lining design parameters, with a FLAC back analysis carried out at the location of each surface extensometer.
Instrumentation provided data on movements ahead of the face, deformation of the lining, once installed, and surface movements for the analysis section. Other input parameters to the model, primarily geotechnical, were then varied until the movements predicted by the model matched those in the field. The back analysis enabled validation of the in-situ stress regime, the geotechnical parameters adopted for design, and the amount of relaxation assumed ahead of the tunnel face, estimated at between 28% and 40%.
Related Files
Typical primary lining design details
Cross-section design at tender stage
