As structures go, it will be a big one, the fourth tallest in London. But in another capacity it is a world beater, the world's highest observation wheel. The British Airways London Eye - otherwise known as the Millennium Wheel - is due to start turning in little more than a year and the programme is daunting.
Design and build contractor Hollandia was appointed in September and allotted just 15 months for its work. 'Design has been going at a breathtaking pace, but in a very controlled way,' says construction manager Mace project director Tim Renwick. Fabrication is under way and the civils preparatory work out to tender (see box).
It will be the biggest Ferris-type wheel in the world, with an overall diameter of 135m to the outside of the passenger capsules. The 32 capsules, each carrying 25 people, will offer a birds' eye view during the half hour rotation. Getting the thought and the concept planning right is essential, says Mace design manager Neil Thompson.
It is a complex structure, and the computer models are enormous, explains Aire Lanser, project leader with Iv-Infragroep which is carrying out the engineering design of the structural steelwork. This is very much a live structure and its dynamic response is a key factor in its design. Issues such as fatigue, wind loading and passenger comfort have to be considered.
Most Ferris wheels are supported from both sides of the hub and spindle whereas this structure is cantilevered out from the river's edge. The steel rim itself has a diameter of 120m and is 8m deep. A spindle surrounded by the hub goes through the centre of the wheel, supported from the bank by a slender A frame. At 22m long, the central spindle would be the height of a seven storey building if stood on its end. Cables anchor the structure back on to dry land, while short towers founded in the river bed will provide restraint against overturning.
Design for wind loading plays a major part in the process. Static and dynamic wind loads have to be considered on the slender structure, and wind buffeting and oscillation can be significant.
The wheel has a steel rim and wind loads are transmitted to the central hub by cables attached to nodes. There are 64 cables, each prestressed to 75t, explains Hollandia's Professor Jacques Berenbak. Berenbak heads Hollandia's design department, and is professor of structural design at Delft University of Technology. 'And the full deadload of the wheel is also hanging on the cables.' The cables at the bottom have more stress than the cables at the top, though most of the wind load is at the top. 'The prestress has to be high enough so that the cables won't slacken under the highest windload,' Berenbak adds.
One of the key factors in wheel analysis has been to look at the buckling capacity of the rim subject to high compressive loads. One of the complications is that the buckling capacity of the rim is low because architecturally it is very slender.
Non-linear analysis was used for stability, plus fatigue analysis. There are two reasons why non-linear analysis is needed, adds Allott & Lomax technical director and independent design checker Dr Allan Mann. 'The behaviour of the members, particularly the cables, is non-linear under increasing stress. Secondly, the mode of failure being investigated, the buckling mode, is dominated by the prestress and is a non-linear effect.'
Several different models have been used. For the dynamic model there are 3,000 beam elements with three nodes and six degrees of freedom at each. The hub and spindle are modelled using thick plate elements, with eight nodes and three degrees of freedom at each - and a total of 30,000 elements.
Much of the modelling is carried out on top end PCs. These are used for the static and linear and preliminary non-linear modelling, while Unix machines have been needed for the bigger, non-linear dynamic models.
'We have been using the two finite element analysis programs because it is such a specialist project,' says Lanser. Strucad was used for initial non-linear analysis and then a non-linear analysis was carried out using Ansys.
Aspects such as vortex shedding, oscillation and fatigue have to be considered. Tuned mass dampers will limit the potential for wind buffeting, and testing starts this week at Dutch research establishment TNO Bouw. This has given specialist advice on wind loading, dynamic response, fatigue, safety philosophy and buckling analysis.
The fatigue designs of the components are also very difficult. 'You can't neglect anything,' adds Lanser. Professor Jaap Wardenier of Delft University of Technology, world leader in tube connections, has been assisting. Nodes in the rim have high fatigue loading, requiring detailed finite element analysis of stress concentrations. The model is very complex, and some 130 load combinations have to be examined, for displacement, stresses, stability and fatigue.
As the structure turns, the stresses in the cables cycle, higher at the bottom and lower at the top. The design life is a nominal 50 years, or about 0.5M cycles. All the connections are designed to be inspected, and if necessary repaired - though this would not be expected during the first five years, and would only be necessary from year 15 on, adds Thompson.
'The scale of this job is so great that secondary effects you might normally ignore, such as dimensional changes and thermal changes also have to be examined,' adds Mann.
Even though the spindle is in essence a cantilever, its size means that additional checks are needed. Dimensional accuracy and distortions are important because it is a tube, with bearings inside. Cross diameter dimensional control and the interface with the bearings and the hub are examined through the computer finite element analysis. The quality of manufacture is very important because of its size and it will be tested in Rotterdam prior to shipping to London.
British, Dutch and Eurocodes are being considered throughout the design, together with good workmanship, Berenbak says. 'Right from the beginning of the job, Arup and ourselves took the view that this is a job for engineers and not people working from codes,' adds Mann. It is no good on a structure like this simply checking for compliance with the codes, he says. Neither BS 5950 nor BS 5400 has any prestressing, casting, or deals with non-linear finite element analysis. 'To some extent we have to work from engineering principles,' he adds.
Passenger comfort during the half hour ride is also important, adds Berenbak. The influence of the wind, the frequency and the damping have to be considered, with the movements and accelerations of the capsules to Beaufort level 6 in operational conditions. Human responses to the motion of rides are complex, but well understood, and the output of the structural analysis can be related to this. 'The final stage of the design is passenger comfort,' says Thompson. This is not going to be a white knuckle ride.