No one realised it at the time, but the day after the water main gas explosion on Saturday June 4 that killed mother of two Philomen Gulland, Raveen Jaduram revealed a terrible truth about the cause.

At the Onehunga site the next day the Watercare chief operations officer told the Herald. "When the pipeline is being emptied of water, air gets in through air valves." He was talking about the air release valves which had been opened along with scour valves to drain down a section of the "Hunua 3" water main. "From somewhere, some source, these gases got in," said Jaduram.

It was the first clue about the bizarre accident that killed Gulland, maimed her colleague Ian Winson and injured several other workers - a tragedy that might have been much worse. Reports of gas leaks in the area had seemingly gone unheeded, and such was the force of the blast that houses in the area shook violently and mud and debris were thrown more than 40 metres.

What Jaduram didn't say was that opening the valves created a suction effect - drawing air (and gas if there had been any leaking nearby) - down into the pipe as the water drained out. The draining process is believed to have taken 7-8 hours, so the section of water main was empty when shift workers arrived about 2am on Saturday morning to begin cutting out a 3m section of the pipe.

Advertisement

"The air valves do offer an opportunity for gas to be drawn into the pipe because during drainage of the water, a partial vacuum would be created in the pipe which would put suction on the air valves," says Auckland University Department of Civil and Environmental Engineering senior lecturer John St. George.

"Any natural gas in the vicinity of the air valve inlet would be sucked in with air."

The suction effect explains how the natural gas used in Vector's city gas supply, which is lighter than air, entered the pipe. "Natural gas rises, it doesn't sink," a Vector spokeswoman said after the explosion. She also claimed there were no gas leaks or gas readings of any significance at the time and that the Vector pipeline sat five metres above Watercare's pipeline, where the explosion happened.

But that information contradicts widespread reports of gas smells in the area - in the week before, the morning before, and in the days following the explosion. In its natural state, natural gas, mostly composed of methane, has no smell. But when used in a reticulated gas supply it has an odorant added - a distinctive scent to let people know when gas is present or leaking.

Fire Service crew didn't take long to identify methane as the gas that caused the blast. But Fire Service area commander Murray Binning, HEB Contractors chief executive Derrick Adams, whose firm was the lead contractor, and Watercare chief executive Mark Ford, all initially said they had no idea of the source of the gas.

Ford told reporters he was unaware of reports that a nearby business had been evacuated on Friday June 3 because of a gas leak.

But a few days later Ford said Watercare was not investigating the possibility of methane coming from a natural source - such as from the decomposition of old landfills in the area. He also admitted staff on site had smelled the odorant that is put into the city gas supply.

The information correlates with what site worker Ramon Milner told reporters. He believes his life may have been saved because he and two others turned up late for work on the day of the explosion. Milner said there had been a strong scent of gas all week and he was surprised the site manager hadn't shut off the valves. There were also concerns, despite prohibitions, about some workers smoking in the trench. And there were other reports of gas smells from business owners and residents nearby.

St George is surprised by the response by staff on site. "Properly trained personnel should be asking why is there gas, if they smell it. With any sort of gas present they should stop work totally until someone competent decides what the gas is, where it is coming from and how much there is in percentage terms," he says.

By 7.15am on the day of the tragedy workers finished cutting and removing a 3m section of pipe in preparation for joining Hunua 3 to the new Hunua 4 water main with a multi-valve bypass pipe. Hunua 4 runs along Victoria St, underneath Hunua 3 at a depth of at least 4m as it crosses Mount Smart Rd into Athens Rd.

The trenching was in basalt rock, in a hardened lava field of uncharted tunnels and caves and in an area riddled with old and new gas, water, sewer and stormwater pipes plus electricity and telecommunications cables.

Watercare would not tell the Herald what length of Hunua 3 was drained down. But talking to water main engineers and looking at the Council's underground services maps it seems likely the pipe was shut off between Moana Ave in Mount Smart Rd and Henderson Place in Church St - a distance of about 1.5km that had been mentioned in a news report.

The pipe gradient is downwards to Church St but undulates in Mays Rd providing a couple of ups and downs along the way. We counted five scour valves and five air release valves along this length of pipe and, walking the distance, found gas main access covers along the entire route.

St George says if gas had got into the water main it would be possible for the air-gas mixture to form layers due to the density difference of the gas (methane) and air.

"The methane is lighter than air and would tend to move towards the top of the pipe," he says. "If the gas was seeping into the pipe it could easily migrate up and form a pocket of methane at a high point."

St George says gas layering was a concern at the Pike River mine during the recovery phase. "This is why, during routine inspections in mines, personnel are required to take readings from the roof and floor using gas detection equipment like a Draeger with a sniffer to sample above their reach."

Layering of the gases provides a possible explanation why, although gas readings were taken when the Hunua 3 pipe was cut, no gas levels were detected.

At around 8pm maintenance planner Philomen Gulland and network engineer Ian Winson prepared to enter the pipe wearing gas detection equipment and harnesses. The explosion occurred 4m into the pipe at a point where the pipe drops down a metre. If a pocket of methane had formed at this point it would have been at the upper level of the 1.9m diameter pipe and possibly above the gas monitors attached to Gulland and Winson.

"I would never be comfortable just checking the air I'm breathing in a tunnel where the roof is higher," says St George. "The standard procedure is you check where you are, you check above to the top and you check below."

Methane forms an explosive mixture in air between 5-15 per cent. "Miners look to get out of a mine when methane levels are between 2 and 2.5 per cent," says St George. What ignited the gas is unclear, although there is speculation it may have been a digital camera or sparks caused by some grinding reportedly occurring outside the pipe.

"Yes, an explosion could be caused by a camera," says St George. "Underground they stop you taking down a digital watch - it doesn't have to be a big spark."

There is also evidence that gas may have still been entering the pipe or trapped there after the explosion.

Several hours after the blast, fire crews detected explosive levels of gas 200m inside the pipe.

And on Sunday June 5 the site was closed after gas was detected 500-700m away in Mays Rd.

Unanswered questions

The Weekend Herald wanted to put a number of questions to Watercare, Vector and the Department of Labour. All refused to comment while the Department's investigation was ongoing.

The questions include:

1 When was the gas supply to the area turned off?
2 How many reports of gas leaks did Vector receive in the days before and after the explosion and how were they dealt with?
3 Did Watercare advise Vector it was draining down a section of the Hunua 3 watermain?
4 What monitoring procedures were in place at the locations of the opened air release valves?
5 What processes are site workers directed to follow when gas is smelled on site?