The most striking aspect of the new build project is the station hall, which is covered by a 3600 m2 roof made from translucent ETFE film cushions resembling successive layers of cloud hovering like mist over a “forest” of up to 17 m high, tree-like, steel-timber columns. The columns support the ETFE roof and the escalators, ramps and pedestrian bridges inside the hall, which is flooded with light through the foil roof and fulfils a number of purposes: a meeting place with circulation areas, waiting zones, SNCF passenger services, shops, places to eat and co-working areas. It serves as a central interchange between long-distance and regional trains, the underground metro lines, a bus station to the east and the adjoining redeveloped quarters of the city. The triangular hall roof and its extension into the artificial landscape at the station concourse allows daylight to penetrate deep into the building, even as far as the metro platforms.
The design for the new railway station at Rennes envisaged a building forming part of an artificial hillside with a roof looking like banks of mist floating over the city. For this reason, the architects decided in favour of a roof surface made of two-layer ETFE film cushions with deliberately fuzzy contours. The film cushions rest on 18 tree-like columns with a total of 80 “branches”.
Tree-like columns made from steel and timber
The roof construction divides into rectangular bays, each supported in their centre by a tree-like column in steel-timber composite construction (Fig. A, E). The distance between the columns is around 20 m. This arrangement allows the “branches” to form an inverted pyramid, which is a very efficient way of transmitting vertical loads. The timber branches carry compression forces (Fig. C, red arrows), while the tension forces are carried by the steel beams (Fig. C, blue arrows), which act as ties below the ETFE cushions.
This structural system has several advantages: the branches are not subject to bending moments and can therefore be attached by simple bolted connections. In addition, it allows the timber components to be slender and their connections to be simple and light. Because the structure is isostatic, temperature fluctuations do not lead to uncontrolled imposed strains (timber and steel have different coefficients of expansion). In addition, the inverted pyramids add to the horizontal stiffness of the roof.
The height, form and angle of the pyramids are irregular and no two connections are identical. The branches, which carry loads up to 50 t, are designed as glued laminated timber beams in oak. The timber elements are attached to the steel columns with welded steel fins and centrally positioned pins.
ETFE film cushion roof
The roof surface consists of four elongated, amorphous bands of ETFE film cushions, around 8 m wide and up to 140 m long, stepped one above the other (Fig. A, B). In most places, these bands form an undulating profile along the length of the building, creating almond-shaped window openings that offer a view from the station hall onto the old town quarter of Rennes.
ETFE films used in cushions are normally attached to a steel frame, which must withstand horizontal forces of up to 500 kg per linear metre – measured after they have been inflated and under roof loadings. For this reason, cushions must have very dimensionally stable sides.
To fulfil the wish of the architect for fuzzy roof edges, MaP3 developed tubular ETFE cushions mounted on a 1 m high edge ladder beam (Fig. D, F). These tubes surround all four roof bays and are highly visible from inside and outside the station. A curving wire grid defines the inner edge of the cushions.
Roof construction
In the area of the main roof, each film cushion strip consists of longitudinal cushions with a conventional cross-section. The span was halved to limit the width of the cushion to 4 m (Fig. F). The long sides of the cushions are attached to a longitudinal intermediate beam and to the two edge ladder beams. Spacers loaded in compression carry the tensile forces created by the stretched cushions. A tie rod under every spacer compensates for the effect of the off-centre forces. Because the forces arising in the film cushions are always in the same direction, irrespective of whether they are caused by wind suction or compressive loadings, the tension rods are subject only to tensile forces and the spacer pieces only to compression forces in every case (Fig. F). The spacers arranged every 7 m also act as bearers for the intermediate beams.
The forces arising in the film cushions
To prevent fatigue caused by alternating loads, the ETFE cushions must remain under tension in all circumstances. The tensile forces within the film are of the order of 100 kg/m. The size of the tension forces on an inflated film can be calculated as the multiple of the air pressure and the radius of curvature.
The recommended radius of a cushion in normal ETFE roofs is around 5 m. A pressure of 250 pascal (25 kg/m²) is enough to achieve this tensile force. The film cushions in the roof of the new railway station in Rennes are designed in accordance with these guidelines, although the tubular cushion radius is only 50 cm – i.e. ten times smaller – which would, in theory, require ten-times the pressure. Tests showed that a pressure of 1700 pascal (170 kg/m2) is adequate for these cushions.
Cushion behaviour where water collects
Rainwater can collect on the ETFE film roof if the air-filled cushions unintentionally deflate or become damaged. The film is elastic and extends under the weight of the water. Depending on the shape of the cushion and its edges, the volume of water can increase until the membrane tears. More water can collect in deep cushions with a small radius of curvature than can collect in shallow cushions. At the same time the cushions must be able to withstand wind loads.
More tests showed that there was an ideal depth for the cushions in this project. They must be deep enough to limit the loads on the film due to wind but also shallow enough to limit the amount of damage potentially caused by collecting water. In Rennes, the calculation of the amount of water collected was done using the ATEx method (Appréciation technique d‘expérimentation), and a test on site was carried out on the most critical cushion (Fig. G, H). The test involved deflating the cushion and filling it up to the edge with water. The results confirmed the calculations and showed that there was no permanent deformation of the film, even after 24 hours.
