Article - Issue 7, February 2001
Creating Eden in Cornwall: the civil and structural design of the Eden Project
Alan C. Jones FIStructE
Adisused clay pit in Cornwall is now home to two massive, state-of-the-art greenhouses, a 2300 seat amphitheatre and extensive visitor facilities. The Eden Project is one of the UK’s major Millennium Projects; it tells the fascinating story of man’s relationship with and dependence on plants. It is a non-profit, charitable, scientific organisation with a commitment to communicate with the public through entertainment, education and involvement.
The architect Jonathan Ball and Tim Smit, who restored Cornwall’s Lost Gardens of Heligan, co-founded the project in 1994. From the start the Eden Project aimed not simply to entertain, but to ‘promote the understanding and responsible management of the vital relationship between plants, people and resources, leading towards a sustainable future for all’ (Mission Statement). The Millennium Commission awarded the project a grant of £37.5 million in 1997, which represents just under half the total cost. The rest has been contributed by numerous organisations and individuals. Already over 500,000 visitors have been to the partially open site and seen the giant greenhouses under construction; this augurs well for the future of the project, which has to fund itself after the official opening in May.
Initially, the brief stated the project must be located in Cornwall, for a number of reasons: the mild climate is unmatched in Europe; it is an area of high unemployment which will benefit from the presence of the project; a number of sites could be identified which were relatively easy to access by road and rail and were close to centres of population. After reviewing the available sites over a two-year period the Bodelva Pit was chosen, located about 5 km east of St Austel. This China Clay pit had been worked for over a century, was approaching the end of its commercial life and was due for decommissioning. Hence, the benefit of land reclamation was added to this list. The pit’s steep, south-facing walls would provide sun and shelter for the ‘global garden’ they would contain.
The Bodelva pit lies within the eastern end of the St Austell granite in a zone of biotite granite. The rock mass encountered on site has been classified into four different grades: Grade 1 which is fresh, unaltered rock through to Grade IV, a highly kaolinised plastic material which can be moulded by hand. The pit covers an area of about 22 hectares and varies in depth from 30 to 70 metres.
We used a digital ground model, based on aerial survey data, to sculpt the pit. The model, created with Microstation and In-roads software, allowed a detailed evaluation of cut and fill operations to take place. As part of the environmental statement for the project, an undertaking was given that the exporting or importing of fill material would be kept to a minimum. This obviously also had important commercial advantages. The landscape architect developed his scheme and provided outlines to be fed into the model. After several iterations, we obtained a balance. Approximately 800,000 cubic metres of material were moved to create the final profiles. No material was exported. The only materials imported were aggregates for concrete, hardstandings and drainage together with organic matter to construct the topsoil and planting medium.
Water management was also important to the long-term stability of the site. The base of the pit is around 30 metres below the natural water table in the area. Stormwater run-off is controlled by bunded swales and ditches interlaced with the car park terraces and landscaped areas. These store water in the short term and augment the retention provided by a series of storage lakes. The stormwater system is designed to contain a 1:100 year storm event. The final pumped outlet into the Bodelva brook to the south of the site is thus strictly regulated to prevent flooding downstream in compliance with Environment Agency flow restrictions.
The principal buildings are the greenhouses, or biomes. They are designed to provide two climate zones, one modelled on the Humid Tropics (HTB) and the other a Warm Temperate (WTB) or Mediterranean climate.
The HTB is up to 110 metres wide, 55 metres high and 240 metres long. Along its length, it rests on all four designated grades of granite, passing from unaltered rock, across the kaolinised clay materials and back to unaltered rock. At the base of the pit it sits on up to 12 metres of fill material. Whilst slightly smaller in form, the WTB (65 x 35 x 150 m) has a similar formation. The Visitor Centre, positioned on the south-west face to give views over to the biomes and external landscaping, sits on a new plateau cut into the side of the pit and extended with fill material reclaimed from the car parking areas.
During the early scheme development period, we realised that the exact profile of the pit would not be known until construction commenced. Initial schemes for the biomes used curved, arched trusses at regular intervals reaching from the base of the pit on to the cliff face. Each truss therefore had a unique profile and associated span. It was critical from both commercial and aesthetic criteria that the pit shape was maintained whenever possible. Wholesale reshaping was not an option. Very little repetition or rationalisation of trusses could be achieved. With the constantly changing topography, as mining continued, it was inevitable that several redesigns would be necessary before the final geometry could be confirmed.
At this point, the design team proposed a radical change to the basic form of the biomes. They developed a series of intersecting domes of varying diameter. The idea was that once the size and relative position of the domes had been determined the shape of the pit became of secondary concern. The structural form of the domes could be confirmed and the intersection line between superstructure and ground determined the position of foundations and extent of cladding. This enabled the team to proceed with design development of the biome envelope before the final survey of the pit was available.
One of the principle criteria in the client’s brief was to maintain the transparency of the envelope at a maximum. In order to achieve this, the selected cladding material had to provide very high levels of light transmission and the structural elements had to be kept to minimum size and number.
After a prolonged study of various geometrical arrangements for a spherical surface, we selected a geodesic arrangement. By adopting the hexagon form derived by Buckminster Fuller, we were able to achieve an even distribution of structural members. By varying the frequency of sub-divisions in each dome, cladding panel sizes can be adjusted to give optimum form and light levels. The original scheme utilised a single layer, unbraced threedimensional space-frame structure with 500 mm diameter circular hollow sections. The envelope was tendered as two packages, steel and cladding. The successful contractor, Mero Gmbh, offered a combined package supplying both frame and cladding. Their proposal incorporated a space truss system, developed over many years. The two-layer structure uses the geometry given in the tender for the outer layer with a combination of hexagons and triangular elements forming a semi-braced inner layer. The combined system is referred to as a hex-tri-hex arrangement. The outer members are 193 mm diameter circular hollow sections with semi-fixity developed at the nodes whilst the inner members are around 114 mm diameter circular hollow sections with pin-ended connections from the Mero system.
The proposed alternative offered considerable reductions in the weight of steel, although fabrication complexity and the number of nodes increased considerably. The number of nodes in the system has a significant effect on cost, as does the size of the cladding panels. The larger the panels, the fewer the number of nodes and generally, the cheaper the cladding. Hence our objective was to develop cladding panels to be as large as possible.
The cladding system
The system we chose to clad the biomes is a pneumatic structure of ‘cushions’. Each cushion is contained within one module of the structure in the form of either a hexagon, pentagon or triangle. On the largest domes, we have used hexagonal cushions up to 10.9 metres across points. The panels are formed from multiple layers of Ethyltetraflouroethylene (ETFE) foil. The foil is extremely thin: each layer is between 50 and 200 µm thick, giving very high levels of light transmission in both the visible (94–97%) and ultraviolet range (83–88%).
The cushions are held in extruded aluminium perimeter frames using a ‘luff’ groove and bolt rope-type detail, known as a keder, derived from sailing and fabric structure technology. The frames are in turn bolted to brackets on the tubular steel structure at regular intervals. Even with such large panels, the whole cladding system only weighs around 15 kg/m2 – a considerable weight saving on the equivalent glass envelope. Thermal insulation values are as good as double glazing and in some instances better than triple glazing when used horizontally.
The ETFE is a modified copolymer which is extruded into a thin film. This means the surface is extremely smooth and when coupled with the antiadhesive properties of the material, gives a self-cleaning surface. Rain washes off any dirt such as bird droppings and the need for regular cleaning is minimal. The material is unaffected by UV light, atmospheric pollution or weathering and extensive testing has shown an anticipated life expectancy in excess of 40 years. At Burgers Zoo in Arnhem, Holland, buildings with foil roofs have been in use as plant houses for over 20 years. The foil panels themselves weigh only up to 50 kg, making replacement a much easier operation than with glass. It is also possible to effect short-term repair in situ using adhesive ETFE tape.
The system is considered to be environmentally friendly. Although the raw ingredients include natural resources such as gas, oil and other minerals, the quantities used are relatively small per square metre of envelope. The manufacturing process does not involve significant use of additives (unlike PVC, for instance) and the foil is recyclable. The inflation units consume energy to maintain the air pressure within the system but the increased light transmission compensates for this in reduced artificial lighting requirements.
The combination of a lightweight steel frame and cladding system (with a combined weight of around 40 kg/m2 of surface area) makes the effect of environmental loads on the structure all the more critical. To achieve the most efficient solution possible, we assessed snow and wind loads in detail in accordance with the current British Standards. We evaluated the consequences of drifting snow accumulating between cushions, or in the valleys between domes. Wind loads were impossible to assess accurately from the standards because of the unique topography of the site and the complex geometrical shape. Therefore we conducted a detailed study using scale models of the development in the wind tunnel at British Maritime Technology Ltd. This demonstrated that the profile of the pit shelters the buildings from the extremes of wind. As the pit is over 60 metres deep and the highest biome is only 50 metres to the apex, the whole development could be considered to be below ground level. The results of the tests supported this, giving design wind pressures well below those initially predicted.
Once we had established the type of cladding and intensity of environmental loads, the design team concentrated on deriving the optimum geometrical arrangement for the spherical structures. The object was to utilise the largest cushion possible in order to maximise light transmission and to minimise cost: large cushions mean fewer connections in the steelwork and reduced length of aluminium framing. The line of intersection between domes posed a particular problem. It was not possible to align the nodes on either side and this was exacerbated as the geometry of each dome had been scaled to give suitable cushion dimensions. We introduced tubular lattice arches to accommodate this and pick up individual node points. The arches are fabricated in segments from curved tubes and site-welded together.
When the pit was handed over for construction, a full topographical survey was performed to confirm the shape of the areas most recently worked. The digital ground model was integrated with the superstructure model to give an intersection line, which formed the setting out for the foundations.
The biomes are supported on a concrete ‘necklace’ which follows the perimeter of the buildings, hugging the contours of the site. The in situ reinforced concrete strip is generally 2.0 metres wide by 1.4 and 1.0 metres deep in the HTB and WTB respectively. We built a three-dimensional analysis model to investigate the interaction between the foundation and the rock mass. We used equivalent spring stiffness to model the varying soil properties of the four grades of granite. Predicted settlements were limited to 25 mm overall, with a maximum differential between adjacent node points of 15 mm. These values were incorporated into the detailed design of the superstructure. At the locations where the foundations passed from the hard granite on to the softer clays, articulation joints were introduced to prevent abrupt changes in level.
During the early foundations design stage the final loads from the steel superstructure were not available from the frame contractor. Anthony Hunt Associates constructed a full threedimensional model of the biomes and performed a preliminary analysis to estimate the foundation loads. These loads were used to design the foundations. Once the final loads were available, a further check was undertaken to ensure the foundation design was acceptable. The model created for the superstructure was then used to perform an independent category III check on the subcontractor’s design.
The design of the ETFE cushions was initially based on theoretical analysis using an iterative solution. It was accepted that the size of the panels was greater than any system previously built by the supplier so a series of physical tests and mock-ups were included in the contract. As a result of these tests, we increased the degree of patterning used to form the cushion profile and omitted the cable net reinforcement in favour of a second top layer of foil. Under negative wind pressures, the principal loads are applied to the top layer of foil and the load capacity can be increased by either providing thicker material or a greater rise in the profile. The former was not recommended as the foil was already 200 µm thick and above this limit, it can become brittle and difficult to weld. As a general rule the cushion profile is based on a maximum of 10% of the span. Above this, the lateral stability of the cushion under transverse loads is unpredictable. Hence the adoption of two separate layers of foil. On the inner layer, which experiences principally snow and positive wind pressures but no transverse loads, the load capacity was achieved by increasing the patterning of the profile to give a rise of 15% of the span under full load conditions. Cable net reinforcement on the inside of the cushions is still required adjacent to the valleys between domes to support the high loads generated by drifting snow. Under normal operating conditions, the cushions will be inflated to a nominal pressure of 250 pascals above atmospheric. In the event of heavy snow, the pressure will be increased to 400 pascals to prevent deflation under sustained load.
Work commenced on site in November 1998 and the first operation was the construction of structural embankments to support the biomes. To create the new ground profile, up to 15 metres of fill was placed in the base of the pit, in an operation described by the client as ‘reversing the mining process’. The principle reason for this was to achieve acceptable gradients on paths and access routes down into the pit and to provide level areas for planting within the landscape. The lower foundations of both biomes rest on top of the fill. Without special measures the settlement under the load would have been far in excess of the limits set for the envelope design. In parallel with work inside the former quarry, we created car parks and visitor access roads on the excavated terraces formed around the pit rim.
We established a site laboratory to identify and monitor the quality of fill material available on site. Fills were graded into two types: ‘acceptable’ fill for the embankments and roads and ‘general’ fill for other areas of soft landscaping. A system of ground anchors was installed, through the foundation strip into the fill, so that the foundation could be preloaded before erection of the envelope commenced. This removed any remaining short-term settlements within the embankments.
Once the embankments had been formed and the consolidation process had commenced, work began on the slope stabilisation and re-profiling of the pit walls. We used a variety of techniques depending on the grade of the underlying granite and the steepness of the slope. For example, all slopes in the Grade III and IV granite up to 50o were achieved by battering the profile and spray-applying a carefully selected mix of grass, shrub seed and fertiliser in an hydrated adhesive gel, to bind the surface. This planting established itself very quickly, taking only days to green up the slopes and becoming a firm carpet of grass in a few weeks.
On completion of the slope stabilisation and when settlement monitoring indicated that the consolidation of the fill had effectively ceased, the construction of the foundation necklace began. The complex geometry made conventional drawing methods insufficient as a form of communication with site. A series of three-dimensional coordinates was provided, giving alignments similar to those used in road construction. These were fed directly into electronic distance-measuring equipment and set out point by point. Despite the horrendous access problems the foundations were completed on programme, in 12 weeks.
This began with the assembly of a vast birdcage scaffold used as temporary support for the erection of the biome frames. Beginning with dome A of the HTB in the west and dome H of the WTB in the east, the space frame was assembled from a kit of individually labelled parts. The primary elements were shipped to site pre-finished by hot dip galvanising. Nodes were prefinished with a zinc rich paint system. Elements were lifted into position initially by mobile crane and on the larger dome in the HTB using a tower crane on piled foundations.
The aluminium cladding frames are bolted to the top boom elements on the ground before erection, leaving only the corner units to be installed at high level. As each dome, and sufficient areas of the adjacent dome to ensure stability, were completed, the birdcage was removed to allow installation of the pneumatic cushions. Cushions are installed using rope access techniques, working from the perimeter up to the apex. Each cushion is flaked (loosely folded) and then stowed into a PVC bag before being winched to its location on the frame. Teams of rope access technicians fit the cushions within the aluminium perimeter frames by slipping lengths of small aluminium extrusions on to the keders and clipping them into the cladding frames. Final weather seals are effected with corner node and extrusion capping pieces, secured by self-drill/self-tapping screws. Once in place the cushions must be inflated immediately to ensure their structural integrity is maintained. Any flapping in windy conditions would inevitably lead to damage.
Future use and maintenance
The buildings have been generally designed for a 50-year life span. What this means in practice has been debated long and hard, as in all projects! The galvanised finish to the steelwork should last at least 30 years without major treatment or repainting, other than for areas of damage. It was for this reason, and because of the relatively low capital cost, that the galvanised finish was chosen. The cladding system will require regular maintenance of seals etc. but the basic components are expected to last longer than the stated life span (although no guarantees can be given on this). Replacement of cushions will follow a similar approach to the installation and permanent man-safe systems are being provided to achieve this. Internal gantries will be hung from the steel frame below the apex of each dome to allow access to the opening roof vents at high level. The building management team will be briefed to undertake regular inspections of the cushions to identify any damage or air leakages. Any repairs that are necessary will need to be executed promptly to ensure the integrity of the envelope.
Cornwall has had an extremely wet Autumn which has provided the best test of our design for rainwater collection and disposal. There have not been any problems with the system which has proved to be able to cope with the worst rainfall since records were started.
At the time of writing the Visitor Centre is complete and open, the biomes are complete bar a few ventilation openings and the planting and internal paths and water features are well under way. The project is due to be completed and open to the public in March 2001 with a formal opening ceremony in May.
Client: The Eden Project
Civil & Structural Engineer: Anthony Hunt Associates, Cirencester
Geotechnical Engineers: John Grimes Partnership
Architect: Nicholas Grimshaw & Partners
Landscape Architects: Land Use Consultants
Mechanical & Electrical Engineers: Ove Arup & Partners
Cost Consultant: Davis Langdon & Everest
Project Manager: Davis Langdon Management
Constructor: McAlpine Joint Venture