LOCALISED BUCKLING in the north wall of Paris Charles de Gaulle Airport Terminal 2E's vault roof is the newest theory for the collapse that killed four travellers on Sunday 23 May.
First analysis of the stricken structure suggested that column head failure was the most likely cause.
In addition to the four fatalities, hundreds were injured when a 24m wide section of the roof covering the airport's newly opened Terminal 2E crashed down almost without warning onto an internal concourse.
The landmark building opened just 11 months ago. It was designed by an in-house architectural and engineering team at operator Aeroports de Paris (ADP).
The terminal is a precast reinforced concrete vault mounted on columns, enclosing a two-storey portal frame concourse structure (see box).
Vault and concourse are structurally independent.
The collapsed section of roof consisted of six contiguous precast arches, each made up of three sections stitched together with insitu concrete.
It was one of only two sections broken into by escalator connections from the main airport building.
Engineers said that from initial inspections it seemed most likely that collapse of the 32m wide concrete, steel and glass roof was triggered by failure either of concrete in the column heads supporting the vault roof, or in the bolts connecting silicone bearings to the vault edge and column heads.
It is known that operator ADP and contractor GTM opted to wrap the top 300mm of 20 columns with carbon fibre external reinforcement in 2002 after cracks were discovered soon after the roof was positioned. This was designed to help contain unexpectedly high bursting forces and prevent the column shoulders shearing off.
It was eventually decided to wrap all 170 columns as a precaution.
But Paris-based British structural engineer Henry Bardsley has proposed a new mode of failure having studied a previously unpublished set of photographs taken just hours after the collapse.
The photographs, he claimed, showed no sign of the column head failure initially suspected by many experts, and indicated to him that the failure began close to or actually between the openings for escalators.
'Steel stiffening sections which should have been in tension have buckled laterally or have been torn away from the concrete shell, ' explained Bardsley, referring to details which he said the photographs revealed.
He said it was likely that although this buckling could have been partially restrained by the footbridge structures, the concrete shell roof would have broken away at the insitu joint and fallen straight down.
His theory is that as the roof fell it would have pushed the southern composite sections off their column head bearings almost intact. This section then sustained serious impact damage when landing around 1m away from the column feet.
Bardsley said he was passed more than 40 pictures taken by an anonymous airport employee before rescue workers started shifting debris in the search for survivors. He has since passed them on to the official enquiry team.
He said the pictures were much more revealing than most so far published, as they were taken of the north side of the structure which, he said, had more 'extensive and significant damage' and 'some important clues are visible'.
But Atkins structural design head Mike Otlet said it was too early to rule out the original column head failure theory, particularly as recent research has shown that the type of column head wrapping used on the project would have been virtually useless (see wrapping box).
'This is a non standard area with complex load paths and there are a number of factors which could have triggered failure, ' said Otlet.
Otlet's view was backed by ICE vice president and director of UK consultant Babtie, Gordon Masterton, and by consultant WSP Group's technical coordinator Stuart Alexander.
'French investigators should take a close look at the means of support, ' said Masterton.
'Eye witness reports suggest that the shell was deforming for about an hour before the collapse which suggests the edges were spreading, ' he added. Passengers saw longitudinal cracks open up in the crown of the vault, prompting evacuation of the terminal.
Once one the column heads had failed, loads would have passed to adjacent columns via longitudinal vault edge beams.
'If there was a systemic problem with the column head, design or construction you would see progressive failure taking place pretty rapidly, ' Otlet cautioned.
Alexander also said that design of the column heads was a likely cause for concern:
The vault edges bear onto the outer corner of the columns.
'You'd normally place loads directly over the centre of the columns, ' he warned, pointing out that the eccentric loading would have aggravated stress regimes normally found in reinforced concrete pier structures.
It is thought that the column heads were subject to considerable movement from the large vault roof, which may have contributed to fatiguerelated failure. The roof itself was designed to cope with 80mm of lateral movement due to wind and snow loading, and with 20mm of thermal movement.
'That would contribute to considerable rocking of the vault back and forth on the column heads, ' said Alexander.
'They would be small movements, but relentless.'