The roofing material and equipment used in the International Convention Centre would be extremely unlikely to be the source of the devastating fire to strike the multi-million dollar project, says an industry body.
The Waterproofing Membrane Association has provided the Herald with a detailed account of how its roofing system works after the bitumen-mix roofing product found itself the focus of speculation following the fire.
It took more than 150 firefighters four days to put the fire out, with 27 million litres of water poured on to the blaze.
The waterproofing industry was left reeling after discussion on the building site highlighted the actions of a single apprentice in the hours after the fire started.
• 'S*** someone's left a blowtorch on': Inside the SkyCity inferno
• SkyCity fire: New pictures shed light on devastating blaze
• SkyCity fire: Young worker's blowtorch at centre of probe
• Convention Centre fire: Who pays for the damage to $700m construction site?
Workers on the site said an apprentice broke off from a smoko break to check a blowtorch left alight where the roofing was being done.
It was said the apprentice was unable to access the area where the roofing work had been done before the alarm was raised over the fire. The story of the apprentice was quickly linked to the blaze as the search for answers as to its cause began.
The Waterproofing Membrane Association senior industry leadership told the Herald the roofing method was governed by a Code of Practice. The organisation regularly met with building industry regulators as part of its professional mandate.
Association chairman Jim Gerbes would not comment directly on the convention centre fire and said he was not personally aware of details.
But he said general information about how the system worked would provide the public with more information on which to base its views.
"It is exceptionally unlikely an installer can get a fire to start while installing the membrane," said the association.
The blowtorches used to heat the bitumen-based membrane would be highly unlikely to allow a flame to be directed on a surface because of its design, the association said.
A "dead man lever" means the blowtorch's accelerated flame died away to a "gentle pilot flame" when not held down. The flame itself was gentle enough to be blown out in wind but did not automatically extinguish.
Even then, said the association, the blowtorch was designed to direct the nozzle upwards if left unattended.
Gerbes said bitumen roofing was once poured hot but development in the 1970s saw bitumen-mix rolls used, with operators using gas-powered torches to melt the product so it would stick to the surface being clad.
He said the industry was tightly controlled through suppliers importing the product and selling it to a network of applicators, most of which belonged to the industry association and through it were governed by a Code of Practice.
The code had an "extensive section" on health and safety and fire safety, he said.
"You cannot just go to the local builder's merchant, obtain the materials and start laying torch-on roofs."
Gerbes said the common surfaces on which the bitumen-mix was laid included plywood.
The first layer of roofing material was laid in strips a metre wide, 10m long and 3mm thick. The blowtorch was used to melt the underside of the strip, then laying it on the surface - plywood in the case of the convention centre - where it cooled and formed a bond.
A second sheet of 4mm was then laid overlapped on the first layer.
Gerbes said the blowtorch was gas-powered, ignited with a flint or lighter after the gas bottle had been turned on. The gas stayed alight at the nozzle with a "gentle pilot flame" until a lever was depressed, forcing gas out and bringing the flame up to operational use.
The lever needed pressure constantly or the flame would revert to the pilot flame.
While the pilot flame would stay on if the torch was not attended, "it is susceptible to being extinguished by wind".
"It is not possible for the flame to reach operational condition if the torch is left unattended because of the dead-man level activation."
Gerbes said the torch was designed with a bent neck or built-in stand and a windshield to protect the pilot flame.
"The result of this is that if the torch is laid down on the roof surface the flame would face upwards at an angle which prevents it impinging directly on the surface."
The roofing company, MPM Waterproofing Services, has refused to comment on speculation around the fire and is waiting for there investigation to finish.
Last week, the Herald asked general manager Andrew Pardington directly about the apprentice speculation. In response, he said "from my understanding that's not what happened".
Asked whether the blowtorch could be left on, he said the equipment used by the company had "built-in safety systems" and the high-heat flame function needed to be done with the operator's control.
"Our people also work in pairs with one watching as a safety officer. Because of the seriousness of the situation we are carrying out our own internal review as I suspect most contractors on site will be."
Nuralite Waterproofing Ltd general manager Shane Clarke, who supplies the industry, said the use of plywood as a base for the roofing material was common across the country.
The roofing system itself had been used on the Supreme Court, Parliament and museums.
"In the 15 years I've been involved in waterproofing and torch-on membranes, I'm not aware of a fire in a new build."
The Herald spoke to a number of people in the industry who described rare fires during renovation projects but none occurring at sites of new buildings.
A spokeswoman for Fletcher Building said it would not answer questions while the investigation was under way.
"As previously disclosed, the product is known as 'torch-on roofing membrane' and is in common use."
SkyCity has also stayed silent on the cause of the fire, saying it would wait on the investigation.
A Fire and Emergency New Zealand spokeswoman said chief fire investigator Peter Wilding was leading the investigation into the fire.
Wilding had been with the fire service since 1987, investigated fires since 1996 and been national manager of fire investigation since 2008.
"Fire and Emergency investigators have finished their work on the site and are now analysing a significant amount of information gathered during the investigation stage," the spokeswoman said.
"Due to the high public interest in this incident, we will publish our findings once completed."
WorkSafe has now launched an inquiry into health and safety practices on the site.