Since September 11, 2001 St Paul's Chapel in Manhattan has become a focal point for a city in mourning.
Unscathed by the collapse of the World Trade Center's twin towers only metres away, the chapel's railings are adorned with T-shirts, baseball caps, flags, banners and flowers in memory of those who died in the terrorist attack.
Every day tourists join New Yorkers in a queue snaking down the street from the tiny church, clutching the free tickets giving them each a few moments on the public viewing platform next door, to pay their respects and to catch a glimpse of recovery work on the WTC site.
Six months on and New York is slowly establishing its new normality. The city's famous shops are open, the lights are shining on Broadway and Times Square and tourists still flock to attractions such as the Empire State Building, although they are now subjected to extensive security checks, searches and demands for photo identification.
Beyond the barriers, work continues around the clock at Ground Zero to recover evidence and human remains from among the dwindling mountain of building debris. While firemen sift through the rubble, engineers are installing tiebacks to stabilise the massive basement slurry wall - the 'Bathtub' - around part of the site.
Peter Rinaldi, engineering programme manager for the Port Authority of New York and New Jersey, says most of the debris has been taken away to the Fresh Kills landfill site on Staten Island.
'Of the estimated 1.5Mt of debris, about 1.1Mt has been removed, ' he says. Rinaldi is on secondment to the New York City Department of Design and Construction (DDC) which is supervising recovery work.
The recovery operation was expected to take a year, but he says it could be finished by June.
'At first, removing debris at and above ground level was OK - there were big pieces which could be handled relatively easily, ' Rinaldi says. 'But the remaining material is the most difficult. It is at the lowest levels and in the most unstable areas.'
It was clear from the start that search and removal of the debris were inextricably linked with wall stabilisation. The Bathtub contained a plaza level and six basement levels including parking, mechanical areas and a train station as well as World Trade Center substructures and foundations for the twin towers, a hotel (WTC 3) and the Customs House (WTC 6).
Built between 1967 and 1968, the 1070m long, 20m high and 900mm thick slurry wall is made up of 6.7m wide panels of 28N/mm 2concrete reinforced with massive cages of 44mm diameter rebar.
Mueser Rutledge Consulting Engineers (MRCE) is acting as geotechnical consultant to site owner the Port Authority. Partner George Tamaro explains that the WTC was built on filled land beyond the original shoreline of Manhattan Island.
Ground conditions are 4.5m of fill overlying up to 9m of organic clay, 3-12m of glacial sands and silts, a thin layer of glacial till and bedrock, mainly mica schist, typically 21m below ground level.
Tiebacks into the bedrock were used to support the wall during basement excavations and detensioned after the basement floors and columns were built. But after the towers' collapse, many internal structures that had helped support the wall were now doing so only with their debris.
'We knew from the start that we would have problems with the stability of the slurry wall, ' Rinaldi says.'Debris was bridging gaps in collapsed material below but once we reached street level we started to lose that bridging. And when we removed more debris we began to take away support for the wall.'
Tamaro was on site the day after the disaster.
His firm was engaged by the DDC to advise on geotechnical aspects of the recovery effort and to design measures to stabilise the basement wall.
Tamaro knows the WTC site intimately because he was involved in construction of the Bathtub in the 1967 when he worked for the Port Authority.
MRCE's first task was to ascertain the state of the wall and the damage to the basement structures and determine the make-up of the debris.
Underground inspections and mapping gave engineers more of an idea of what they were dealing with.
'The wall is doing what it should be doing under the circumstances, ' says Tamaro.'It has been hit, crushed and scorched - it has really taken a beating but is still doing its job.'
Engineers usually rely on meticulous planning and careful calculation before doing the work, but this was not an option at Ground Zero, he explains.
'Most of the decisions were on gut instinct rather than by numbers. There were a lot of judgement calls.' On the spot decisions were often backed up with calculations and observed behaviour at a later date, he says.
'For example, there was a tremendous amount of pressure to put machinery within 60ft [18m] of the wall.'
Tamaro had to fight to prevent this, no matter how great the temptation for contractors to get cranes nearer debris to speed up recovery. 'The machines would have been in the hole if they had been any closer.'
Co-operation between consultants, contractors and rescue personnel and data from instruments around the site, along with years of 'irreplaceable experience'have helped confirm early decisions.
Once on site, the team quickly came up with a 'three-pronged programme' says Rinaldi. This comprises dewatering, monitoring and stabilisation of the slurry wall, once debris is removed, by installing replacement tiebacks.
Dewatering, which started at the beginning of October, was designed to relieve hydrostatic pressure and lower groundwater levels along the west (West Street) and south walls (Liberty Street) where there was most concern about wall movements.
About 40 wells are on line, discharging about 760 litres a minute. Wells consist of submersible pumps in slotted pipes placed in 305mm diameter boreholes filled with sand and keyed into the bedrock. Dewatering has significantly reduced water pressures next to the slurry wall and lowered groundwater levels to between 7.6m and 12.2m along Liberty Street and West Street.
The monitoring programme records wall movements and groundwater levels around the wall.
Some 17 upper and lower monitoring wells, 14 inclinometers and 60 optical survey points have been installed to date and more will follow, Rinaldi says.
The upper monitoring wells are typically 6m deep and installed through the fill and into the organic clay to monitor shallow perched groundwater. Lower monitoring wells are installed and screened at the top of the rock or in some cases at the top of the dense till to monitor groundwater below the organic clay.
Movement was monitored by inclinometers installed 1-4m from the wall, keyed a minimum of 3m into rock, and optical survey monitoring points were chiselled into the top of the slurry wall.
With the vast numbers of people and amount of equipment on the site, finding safe instrument locations was difficult. Despite the team's best efforts, including putting 1.2m diameter, 900mm high and 100mm thick concrete manhole risers around installations, about a quarter of the instruments have been damaged.
Drilling and grouting the tiebacks also disturbed the inclinometers and some of the monitoring and dewatering wells were accidentally grouted up.
Installation of the replacement tiebacks began at Liberty Street. It was here that one of the most worrying moments occurred.
On October 7, both Rinaldi and Tamaro were alerted to massive movement of a section of the slurry wall at Liberty Street.
'We were about to start tieback installation here when I got a call at 7.30am to tell me that a crack had opened up behind the wall in the middle of the night, ' Rinaldi recalls.
The 60m long tension crack lay 8m behind the wall and was up to 100mm wide at its centre. After visual inspection it was clear that it was a 'classic' earth pressure wedge failure, with about 50mm of settlement behind the wall.
He says the main problem was that debris lying against wall was of unknown composition and density. Visual inspection revealed the collapsing south tower had ripped out four of the basement slabs at Liberty Street. This, combined with a large void in the debris, left unsupported a 46m long, up to 14m high section of the wall, equivalent to seven panels.
Tamaro immediately decided to start emergency backfilling. 'Some of the intermediate slabs were still there and while it [the wall] was stable I didn't want to backfill, but the minute it did move, we did.'
Trucks, dumpers and tracked conveyors were used to rapidly place fill through the debris and holes punched out of the slab remnants. Four days of backfilling slowed the movement and a further three days saw temporary stabilisation, but by then the wall had moved more than 300mm.
Emergency tieback installation then began in earnest. Although they were in relatively good condition, it was not possible to re-tension the original anchors. 'When they were cut they were sucked back into the ground, ' Tamaro says. 'We tried to drill around to get to them but we couldn't advance the core barrel very far.'
One of the main challenges was where to install the replacements. 'Ideally they should go where the original ones were but obviously they could not, ' Tamaro says.
The original plan was to install a pair of tiebacks either side of a panel joint - 'we were worried that the very large loadings would cause differential movement between the panels, ' Rinaldi explains.
But after the first two pairs were put in it was found that the differential movement was small - 'so we decided that we could install them anywhere' The new tiebacks typically comprise 22, 15mm diameter steel strands, installed by contractor Nicholson Construction Company at 3.7m centres along the wall at 45infinity from horizontal. Basement slab remnants are removed before a 300mm hole is cored through the wall. A tieback trumpet is then installed and the steel cased 180mm diameter borehole for the tieback drilled down to bedrock, with a minimum 0.6m long, 150mm diameter rock socket. After the strands are inserted the hole is grouted over its full length, with the casing left in.
Tiebacks are installed by three different types of drilling equipment. Where debris and basement slabs have allowed, drilling has been carried out by a combination of large tracked rigs and smaller, lightweight tracked rigs. Hanging leads suspended from cranes have been used where the ground is less stable.
Tamaro says an average of eight tiebacks are installed a day with up to eight machines working at once.
The tiebacks are proof tested to 800kips (3559kN), 133% of the design load, and locked off at 600kips (2669kN). At any one time between 20 and 30 tiebacks are awaiting testing while the grout is curing, which takes about three days. A maximum of 14 tiebacks have been tested in one eight-hour shift.
So far, only one of the tiebacks has failed to hold the test load, but this was in an area where the bedrock is known to contain anomalies.
Tamaro says the tiebacks were designed for ductile or bending failure of the wall rather than tension failure of the anchors to ensure that the wall did not fail catastrophically.
He adds that this was another example of making a decision and then backing it up later. 'We looked at the old loads and then made sure that the new tiebacks would work using measured performance - we have since gone back and done the calculations to confirm it.'
Tamaro estimates that about 1150 anchors will be needed to stabilise all the wall sections, although this is 'a moving target - we won't be sure until the slabs are removed.' More than 500 had been installed by the end of February.
The original number of replacement tiebacks has been reduced by about one level across the site, Tamaro says.'We are down to four levels where five were planned and five where six were planned.' This was possible because of the increased knowledge of anchor behaviour and because the new tiebacks have a 15% higher capacity than the original ones, he explains.
Liberty Street moved another 25mm while the tiebacks were installed. Tensioning actually pulled back the top 6m of the wall by some 60mm.
Work then moved to West Street. Rinaldi says: 'While the wall didn't move here, there was the greatest potential liability from failure because of the amount of groundwater behind and the ability to recharge.'
When Ground Engineering visited, excavation was down about 80% to subgrade level at the Liberty Street end and down to 50% in the centre.
Buildings that were damaged beyond repair have been demolished to street level or below - the last of the above ground remnants of WTC 6 were removed in late December. Excavation has since reached bottom at several locations.
The next challenge is in the northern part of the site where almost all of the basement slabs are intact. Tieback installation will have to work in and around these structures - they are not being removed just yet, Rinaldi says.
Meanwhile, a new prefabricated steel access ramp is being built, which will allow the 'earth' ramps to be removed. These were built on debris which must be excavated and searched.
As to Ground Zero's future, no decision has been made.Whether it is left as a memorial to those who died or redeveloped as public open space or commercial properties remains to be seen. It will be up to New Yorkers to decide.