By David Richards, Ken Rush Associates, and Michael Turner, Applied Geotechnical Engineering.
Permanent ground anchors are an efficient and cost effective way of providing restraint for retaining structures, involving minimum disruption at surface level during construction. The 1970s and 1980s saw a marked growth in the use of ground anchored systems with design lives in excess of 60 years. However, the next cycle of redevelopment on sites where they were fi rst used is now beginning, and the implications and constraints imposed by these anchors are becoming apparent.
This paper describes the design and construction of piled foundations which were installed at Chandlers Wharf, Erith, Kent in the proximity of permanent ground anchors as part of a residential development of a derelict industrial site alongside the River Thames.
An initial desk study located not only original drawings and records but also traced the anchor designer.
This helped to develop a detailed understanding of the particular anchors used at Erith and the constraints these would impose on the design of new piling. Aspects of the construction works are also described, including monitoring and testing of the ground anchors.
Introduction The Chandlers Wharf site at Erith was once dominated by industry, having been used variously in the past for shipbuilding, timber saw mills and warehousing. However these industries declined, and by the late 1990s the site was semi-derelict (Figure 1).
As part of the 1970s programme of tidal fl ood prevention along the banks of the River Thames, the site was provided with a new sheet piled river wall, tied back with permanent ground anchors to minimise disturbance and disruption to the site owners. This was surmounted by a 2.5m high reinforced concrete fl ood defence wall.
However, a generation on, and with the industrial buildings now derelict, these anchors presented a major obstacle to redevelopment.
In the 1990s the site was identified as a prime location for a residential development with waterfront views.
The proposed scheme comprised some 235 fl ats in nine blocks and 350 parking spaces at ground level.
As might be expected, the preference of the developer was for the maximum number of fl ats to have river views. An outline plan of the proposed position and orientation of the apartments is shown on Figure 2. This also shows the location of the fl ood defence wall and its ground anchors.
Relocating the buildings back from the river, and away from the footprint of the anchors, was not deemed to be a viable option by the developer, since the amenity value of the river would be lost and there would be a signifi cant reduction in density.
It was therefore necessary to design and construct a suitable foundation system for these six to eight storey buildings, which would be located above highly stressed permanent ground anchors.
This paper describes the appraisal of the impact of the ground anchors on the proposed development and the design and construction of the piling and associated substructures around the constraints imposed.
Solving these problems ultimately allowed the developer's proposals to be realised.
Desk study and initial appraisal At the design stage it was clearly vital to understand as much as possible about the design, details and locations of the ground anchors, so that the constraints they would impose on foundation design could be established.
The initial desk study located copies of some of the original drawings, calculations, anchor test results and 'as-built' schedules in the archives of the Environment Agency, which had taken over the responsibility for the Thames fl ood defences from the Greater London Council.
Although the original works had been undertaken only 20 years previously, it was almost by chance that even this relatively limited amount of information still existed, since much data had been lost during this transfer of responsibility.
Included in the available records were schedules of the design inclination, load and anticipated length of each anchor. In a few cases an angle of skew on plan was also noted.
This information appeared to correlate well with an on-site survey of the sheet piled river wall and the anchor heads. This allowed ground anchor positions to be plotted accurately on the site plan relative to the proposed buildings and was a vital fi rst step in ensuring that piles were positioned to avoid damaging the anchors.
Without a viable means of tracing the as-built anchor alignments (by some geophysical means, for instance), it was realised that the anticipated position of an anchor in the ground would have to be based on the available records, possibly confi rmed only by visible evidence at the head and with appropriate allowances made for construction tolerances.
However, without knowledge of the method of installation or even the form of construction of the anchors, such allowances might have been merely guesswork.
During this desk study phase, the original anchor designer (and co-author) was traced. In addition to specialist knowledge and advice, this gave access to further records and information regarding the original method of construction and installation.
Ground conditions Some original ground investigation data was available, and this was supplemented by additional boreholes and probes.
Ground conditions were fairly typical of a riverfront site over this part of the Thames. The site was underlain by a variable thickness of made ground, overlying soft alluvial silts and clays to a combined depth of between 8m and 11m below ground level.
Beneath this was a relatively thin band of Flood Plain Gravel, overlying Chalk. The gravel was up to 3m thick in places, but was absent at the eastern (downstream) end of the site, where the chalk correspondingly rose in level. A typical east-west geological section through the site is shown in Figure 3.
The information enabled an estimated cross-section of each ground anchor to be plotted to verify existing records, particularly regarding overall lengths, depths and lengths of anchorage into the chalk.
Evaluation of the anchored wall The existing fl od defence wall comprises an inverted T-shaped reinforced concrete wall approximately 2.5m high. From the available records it was established that the landward toe is supported on 350mm diameter cased piles, typically at 2.9m centres.
The outer toe bears on to Frodingham 3N sheet piles approximately 16m long and tied back with ground anchors at 3.85m centres, raking, typically, at 40¦ to the horizontal. A typical cross-section through the anchored structure is shown on Figure 4.
Although no contemporary detailed analysis of the sheet piled wall and anchor system had survived, original calculations for the design of the upper tee-shaped reinforced concrete river wall helped in assessing the design load cases.
These included a 'surcharge' from within the site, active pressures from the retained soils, and groundwater pressure in the particular case of a surge tide overtopping the wall, followed by the succeeding low tide: when the differential in water levels either side of the wall would be at a maximum.
A design 'back analysis' confirmed the provision of a horizontal component of restraint of 100kN/ m, as originally designed and provided by the anchors, was within the correct range to meet this design case and that this degree of restraint would therefore need to be maintained.
The signifi cance of this analysis was that it confi rmed there was no potential design redundancy in the anchor system. Hence, damaged anchors would have to be replaced, or their loss compensated for in some way.
Description of the anchor system The ground anchor system was described as a fully or double-protected system, and was of the resin encapsulated steel strand type (Turner ).
With this system, the anchor tendon is formed from low relaxation prestressing strand, with its lower (distal) end encased within a high strength polyester resin, reinforced with a steel spiral, to form the tendon bond length.
This capsule or encapsulation transferred anchor loads into the founding chalk strata along the grouted fi ed anchor length. The upper free tendon length of the strands was covered within individual plastic sheaths. Figure 5 shows a typical section through a ground anchor and head assembly of the type used at Erith.
The development of permanent double corrosion-protected anchors in the UK was given a great impetus by the demanding technical requirements associated with the construction of the Thames Barrier and downstream fl od prevention and bank raising works in the mid to late 1970s.
The need to meet the exacting technical specifications developed by the Greater London Council spurred innovation and gave added importance to developing a suitable British Standard for the systems being developed. Terminology subsequently adopted by BS DD81 and BS8081 [2 and 3] was developed from these designs. Figure 6 defines anchor terminology used below.
At Erith, most anchors had a specifi ed working load of 550kN, and comprised fi e plastic coated 15.2mm diameter drawn or 'Dyform' strands. These were encased in a 3.5m long, 70mm diameter resin capsule.
From the available calculations, the design fi xed anchor length for such a load was 10.5m, and the total penetration into the chalk was typically 16.5m (ie there was an initial penetration of 6m into the chalk before forming the design fi xed anchor length of the ground anchor).
From Figures 5 and 6, it can be seen that the load applied at the sheet piled wall via the anchor head is transferred through the debonded free tendon length into the steelreinforced resin capsule.
Load is transferred into the capsule by bond between the steel strand and resin. It can be envisaged that transfer of the applied load into the grout filled fi xed anchor length, and then into the surrounding ground will take place from the upper (proximal) end of the capsule.
Hence, it would be expected that the most highly stressed region of the capsule would be its proximal end, and that the applied jacking force in the anchor would be shed progressively downwards from this point.
At the same time, the applied force would be transferred from the encapsulation into the surrounding cement grout fi ling the drill hole and forming the fixed anchor length. The most highly stressed zone of the fi ed anchor grout would therefore also be expected to be adjacent to the proximal end of the encapsulation.
Original tests at the development stage of the ground anchor system  demonstrated that the bond between the resin and the embedded steel strands could normally be fully mobilised over a 600mm test length.
For evaluation purposes, therefore, for the Erith anchors, the authors assumed the most highly stressed zone within the fixed anchor length would extend from about 1m below the proximal end of the encapsulation to about 1m above, with stresses falling to zero on either side over the design fixed anchor length. This concept is summarised in Figure 7.
Extending the concept further, for a typical 550kN anchor, a design fi xed anchor length of 10.5m divided by the design Factor of Safety (of 3) would imply that a 3.5m length would be capable of transferring most of the load into the chalk.
A 3.5m long zone centred on the proximal end of the encapsulation could therefore be expected to support a large proportion of the anchor load and be highly stressed.
This analysis suggested that piling near to this zone would consequently need special consideration.
Constraints on piling imposed by the ground anchors Having established the theoretical positions of the anchors from records, cross-checked these with the topographical survey of the visible anchor heads and gained an understanding of how the anchors were working, the constraints on piling, due to the post-tensioned anchors, could then be evaluated.
Evaluation of design cases The first aim of the evaluation was to identify those areas where piles could be located with some degree of confi ence, and those where piling would be prohibited due to an unacceptable risk of physically damaging anchors with the piling equipment.
From a consideration of the design and construction features of the ground anchors, three distinct cases could be envisaged:
Case 1: Areas suffi ciently distanced from the anchors that no limitations would need to be imposed.
Case 2: 'No-pile zones' where piling would not be permitted at all.
Case 3: Areas near to anchors where piling would be permitted but subject to certain limitations.
The first and most obvious stage was to try to prevent direct physical damage arising from piles cutting through anchors. It was known that each anchor was subject to specified setting out and construction tolerances when installed, and the initial step was to determine realistic assumptions for the likely deviations, as built, from the theoretical alignment.
If values of likely deviation were assumed to be too conservative, the site might become virtually 'sterilised' and undevelopable. Unrealistically optimistic values of deviation would greatly increase the risk of damaging anchors.
Likely positional tolerances for the ground anchors As an example, allowable setting up tolerances in typical specification documents (eg  and ) would be around ¦2.5¦, with a permissible deviation due to installation/drilling effects of around 2¦. Hence, the gross permissible tolerance for the installed anchor might be as large as 4.5¦.
The anchors were installed using a Hymac-mounted hydraulic rotary drill, with the mast bolted to the sheet piles during drilling. Vertical alignment was set out by inclinometer.
Therefore it was postulated that for both vertical and horizontal alignment, setting up accuracy would be within 1¦to 2¦ of that originally specified.
In addition to these 'setting-up' tolerances, allowance would have to be made for possible deviations arising from the drilling process.
It was felt that two references gave some guidance on this aspect. On a ground anchor project on the A55 trunk road in North Wales , the as installed positions of Odex temporarily cased ground anchor drillholes were monitored by down hole surveying techniques.
This demonstrated that most anchors had been installed within a positional tolerance of 1¦ of the theoretical location. In the second case study, at the British Library in London , average positional accuracies of within 2¦ of theoretical had been recorded. These latter anchors had been installed using a more flexible, continuous fl ight auger, drilling system.
It was known that a relatively stiff duplex drilling system  had been used for the Erith anchors, so that deviations due to drilling could be expected to have been relatively well controlled.
The sum of these possible deviations of an anchor from its theoretical position can be visualised as a cone with its axis along the intended alignment of the anchor, and its apex at the point of entry of the anchor (a 'tolerance cone'), Figure 8.
It was decided a tolerance cone with an apex angle of 5¦ would be appropriate for this site and this value was used to defi ne the likely extremes of position of any particular anchor, both in plan and elevation.
It was realised that this was a fairly tightly drawn value, given the potential scope for deviation within the system. It was essentially a pragmatic view and carried with it the understanding that it had not reduced to zero the risk of the piling rig striking an anchor. The view was taken, however, that this was a reasonable approach given the enhanced knowledge of the layout, design and method of construction of the anchors.
Considerations on piling tolerances In addition to uncertainty in the precise location of the anchors, installation tolerances of the new piles also had to be considered.
Typical piling tolerances, as outlined in the ICE Piling Specification, SPERW, for instance , allow a positional tolerance at ground level of ¦75mm and a verticality tolerance of 1 in 75.
However it was judged that for a pile to be out of position by 75mm and installed at 1 in 75 in the same direction was too conservative an assumption, in view of other tolerances being considered.
By allowing for a notional positional tolerance of ¦50mm, combined with a verticality tolerance of 1 in 100, a 'tolerance cone' could be drawn, in a similar manner to that visualised for the ground anchors.
To simplify the pile setting out and evaluation process a 'tolerance cylinder' of radius 300mm greater than the pile was adopted, which was taken to be applicable to all piles (Figure 8).
Where adjacent to anchors, piles were therefore positioned such that the tolerance cylinder of a pile did not encroach on the 'no-pile' zone of the anchor. Although not a contractual tolerance, the piling contractor was made fully aware of the constraints of the site, and rose well to the challenge of achieving these enhanced targets.
Proximity of a pile to an anchor Having arrived at an evaluation of the size of the design tolerance cone and tolerance cylinder, a view had to be taken on how close the piling tools could be allowed to approach the envelope of the tolerance cone of an anchor. The decision taken was that:
l Over the free anchor length, it was deemed necessary only to ensure that a pile did not physically touch or damage the tendon or its sheath.
A further tolerance of 150mm was therefore added to the tolerance cone of the anchors to define the 'no pile' zone.
l Over the fi ed anchor length, however, where the anchor force is dissipated into the surrounding ground, a greater separation would be necessary, particularly around the highly stressed central zone near the proximal end of the anchor encapsulation, to reduce the risk of unlocking existing stresses in the chalk.
For this site, given that both piles and anchors were founded in chalk, it was judged that a minimum separation of 450mm from the centreline of the anchor (assuming it to be located on the edge of the tolerance cone) to the face of the tolerance cylinder of the pile would generally be sufficient over the fixed anchor length. In the 'highly stressed' central zone this was increased to 1,250mm. This latter value was a rounded down value equivalent to approximately three times the minimum separation value.
These positional constraints then allowed the three piling cases to be plotted in section (Figure 9) and on plan (Figure 10) for the footprint of each building on the site.
Using these constraint drawings, detailed drawings for each structure could be prepared and piles could be positioned graphically to suit the structure above. Figure 11 shows how the structure specific drawings also incorporated details of any limitations on toe depth as necessary.
Constraints on pile type In addition to reducing the risk of direct damage to the anchors by careful positioning of piles, the risks of indirect damage which might be caused by installation of the pile were also considered.
Normally a driven pile would be well suited to the type of ground conditions on this site. However, as vibrations from installation could damage the anchors or disturb their bond within the chalk, it was considered that a replacement type bored pile would be used.
However, by implication, the act of forming a bore allows the ground to relax to some degree, before the piling concrete restores support after initial set.
To minimise the risk of removing or reducing support to the anchor over its fixed length, it was therefore considered that the piling system for this site would have to:
a) form the bore quickly b) provide support to the bore during construction c) restore support quickly by placing the pile concrete as soon as practicable after boring was complete.
For these reasons, continuous flight auger (CFA) piles were regarded as most appropriate. Even this system, however, could result in a short term loss of support within the ground until the concrete started to gain strength.
Therefore, as a precaution, it was specified that not more than one pile should be constructed next to the fixed anchor length of any one anchor in any one day. This would allow for initial stiffening of the concrete in a pile, helping to restore support to the chalk, before opening up the next bore.
It was judged that by using a rapid installation type of pile to maintain ground support and by ensuring that the toe or shaft of the piles was at least 1,250mm away from the tolerance cone in the vicinity of the highly stressed zone of the anchor, any effects of constructing a pile adjacent to a ground anchor would be minimised, or, at the least, controllable and capable of being assessed by monitoring the behaviour of the relevant ground anchors.
Safety Due to the high level of stored energy in the strands, explosive failure was considered possible if the pile auger came into contact with an anchor. The designer's risk assessment therefore recommended installation of energy absorbent covers over the anchor heads during piling to contain the strand and cap in the event of sudden failure.
A condition was also imposed that, should failure of an anchor knowingly occur, no other piling should proceed in the adjacent area until a replacement had been installed. This was to guard against a 'progressive' failure if neighbouring anchors were also cut and the remaining adjacent anchors became overloaded as a result.
Design of substructures and superstructures Redevelopment of the site comprised construction of nine new blocks of flats ranging from three to eight storeys (Figure 2).
These were of two generic types:
linear blocks (A, E, F and G) of load bearing masonry and precast concrete fl oors, which ranged from three to five storeys, and 'tower' blocks (C, D, H, I and J) which were of fi ve to eight storeys and used in situ concrete frame construction.
All of the tower blocks were located either wholly or partly over the footprints of the ground anchors.
The blocks were orientated at typically 30¦ to 45¦ to the flood defence wall to maximise the benefits and amenity value of the river views.
However, there was a requirement for 350 car parking spaces at ground level. An effi cient layout was required to fit all these spaces on to the site and still allow for areas of soft landscaping.
This meant that at ground level each tower block incorporated varying amounts of car parking and vehicular circulation space in addition to bin stores, plant rooms, lifts, stairs and entrance lobbies.
To achieve this, a first floor 'transfer structure' comprising heavy downstand beams was used to adjust the grid of columns above (located to suit the fl at layouts) to the column grid at ground level which generated the most efficient car parking and circulation.
This optimum car parking column grid in turn had to be superimposed on to the piling constraints plans which had been developed from the parameters described previously.
In each case this meant that a second level of transfer structure was required over at least part of the footprint of each tower block to transfer loads from the ground level columns to locations where piles would be permitted.
However, it became clear at the detail design stage that in some areas beneath several blocks it would prove impracticable to fit sufficient piles within the permitted zones to support the calculated loads. In such cases, it was decided that selected anchors would have to be removed to provide larger 'windows' for the piling works.
These anchors would be replaced by new shallow deadman tie bars, which allowed the corresponding anchors to be destressed and decommissioned, thereby removing all restrictions relating to piling adjacent to them.
Replacement ties bars and deadman anchors Replacement ties comprised conventionally designed M64 Macalloy bars with pvc sheathing and a denso tape outer wrap. At the river wall, the bar passed through the in-pan of the sheet pile adjacent to the existing anchor head and at the same level so that loads would be distributed horizontally by the existing walings on the inside of the wall.
The deadman anchor itself comprised LX8 sheet piles about 3m long, which were driven down to a level beneath the underside of the proposed substructure. A typical confi guration of such replacement ties is shown on Figure 11.
Tie bars were laid on to a wellcompacted engineered fill comprising granular material laid on to a geotextile separation membrane.
This was not only to provide a firm support to the ties to reduce risks of deformation when backfilled, but also to provide a well-drained and fi m working platform for subsequent piling and other construction.
A view of the site during the construction of such replacement ties is shown on Figure 12.
These tie-bars were installed ahead of piling, and were prestressed to a load similar to that in the corresponding anchor which it would replace. These loads were determined from the pre-piling checklifts described in the next section.
Prestressing of the completed anchors and tie bars, however, also had to be integrated carefully within the piling programme. This was due to the lateral loads which would be imposed on bearing piles near to the new deadman anchors, once the ties had been stressed. This is described further under the piling section. Once the new ties had been tensioned to their design loads, the redundant anchors could then be decommissioned.
Ground anchor monitoring Although the above analysis and strategy for positioning of new piles had been agreed with the Environment Agency (EA), it was also deemed to be prudent that the ground anchors should be inspected and monitored both before and during the project.
As no records of any maintenance since their original installation were available, this served not only to provide the developer and the EA with information about their condition and retention of prestress, but was also intended to demonstrate that the wall and anchors were not being adversely affected by the piling and associated works.
The initial monitoring process comprised load checks (checklifting) of every available anchor before piling began, together with an inspection and record of the condition of the anchor head assemblies and caps.
The anchors were checklifted after the completion of piling, and further selective checks were undertaken about three months later. In addition, the river wall itself was monitored by regular theodolite surveys. The aim was to use this as an early warning system to detect abnormal movements which might indicate overload, distress or failure of a ground anchor.