Tank 11 is burning at the Milford Haven refinery in the United Kingdom. - Photo by Barry Browning.

Tank 11 is burning at the Milford Haven refinery in the United Kingdom.

Photo by Barry Browning.

It is a record that any refinery would be happy to relinquish. Before the history-making storage tank fire in Norco, LA, in June 2001 the largest such fire on record happened in August 1983 at a South Wales, U.K., refinery where a 250-foot-diameter tank burned.

Unlike Norco, the Milford Haven fire involved crude oil, not gasoline. With crude oil comes the risk of boilover. And, whereas Norco is considered a successful extinguishment, Milford Haven is not.

"All we had were small capacity foam monitors compared to the size monitors available today," said Barry Browning, a private consultant. He is also a former refinery fire chief who was on hand for the Milford Haven fire.

The LASTFIRE Project, a study of the risks associated with large diameter open-top floating roof storage tanks, covered Milford Haven in detail in its 1997 report. Tank 011 was a 256-foot diameter open-up floating roof storage tank with more than 250 million barrels of light North Sea crude. Inside the tank, the single-skin roof floated at 15 feet below its 65-foot-high maximum.

Unbeknownst to the operators, crude had seeped through cracks in that single skin. Although not cited in the LASTFIRE report, speculation about the cause centered on constant flexing of the metal due to wave action stirred by high winds.

Ignition occurred between 10:45 a.m. and 10:53 a.m. on Tuesday, August 30, probably from incandescent particles from a nearby flare stack. The tank had no automatic fire detection or fixed firefighting system. By the time responders arrived at 11:05 a.m., 50 percent of the roof was covered in flames. As it spread, the fire destroyed the rubber seal between the roof and the tank wall, releasing more flammable vapor.

The first response consisted of a 105-foot hydraulic platform, a 4,500-gallon capacity foam tanker, a 5,000 gpm foam monitor, four water lines and four firefighters. This was rapidly upgraded to a call for 20 pumps and five hydraulic platforms. Within 15 minutes, it was possible to determine the tank's ullage from the level of scorching on the shell plates.

With fire spreading over the entire roof within the first hour, firefighters used a hydraulic platform to deliver foam on the flames but discontinued the effort when it became obvious a major foam attack was needed. By mid-afternoon there were more than 150 responders, 26 pumps, seven foam tenders, six hydraulic platforms and four other special appliances on site.

Among the responders was Browning, who that same year had become fire chief at Esso's Fawley refinery. The shortest route from Fawley to Milford Haven was almost a five-hour drive.

"I invited myself," Browning said. "And they were delighted to receive me."

Throughout the afternoon, operators pumped fuel from Tank 011 and the two tanks 200 feet downwind of the fire. One contained 4,500 barrels of vacuum gas oil and the other held 2,800 barrels of atmospheric fuel oil. As for 011, the crude oil burn-off rate was estimated at 300 tons per hour, compared to a pump-out rate of 1,700 tons per hour.

A tank neighboring Tank 11 after the massive boilover that swept the refinery. - Photo by Barry Browning.

A tank neighboring Tank 11 after the massive boilover that swept the refinery.

Photo by Barry Browning.

Obtaining enough foam to attempt extinguishment became the major stumbling block in the Milford Haven operation, Browning said. Fire officials estimated that at least 50,000 gallons of foam concentrate would be needed.

"It was agreed that the initial foam attack would be at midnight Wednesday," Browning said. "Meanwhile, adequate foam stocks would be accumulated. But obtaining foam from the Angus and Chubb inventory in the U.K. involved very long travel distances. There were three other refineries close by that supplied foam concentrate, but as midnight drew near we were still 20,000 gallons short of enough concentrate to attempt a foam attack."

At 11:30 P.M., 12½ hours after ignition, firefighters tested a single foam monitor against the fire. It produced an encouraging split in the flame but also cause oil to slop into the dike. Within minutes and without warning a boilover occurred. Flames reached 3,000 feet high and covered four acres of the 295 foot by 590-foot dike.

"Everybody ran for their lives," Browning said.

The firefighters were aware that the risk of boilover existed, Browning said. However, it was believed that a boilover could be safely anticipated by warning signs such as a change in the smoke pattern, rising steam or rumbling inside the tank.

"All that went out the window I can assure you," Browning said. "There was no indication at all. One minute it was burning as one would expect, then woof and burning oil was being ejected from the top of the tank and into the dike."

Surprisingly, the boilover claimed no lives. Browning attributes this to the large, 20-foot-deep dike surrounding the tank. Also, the tank was relatively isolated, nearly 3,000 feet from the refinery's nearest process unit.

"There were a few minor injuries and burns and a few fire trucks got a bit blistered," Browning said. "It destroyed much of the fire hose. The water inside it boiled and split the hose."

Radiant heat from the first boilover ignited the thermal cladding on the downwind tank containing atmospheric fuel oil. After a second boilover in 011 about two hours later, the downwind tank buckled and split. Escaping vapors ignited but since the tanks threatened by 011 had already been foamed the fire died quickly.

At 8 a.m. Thursday, 44½ hours into the fire, a decision was made to attack again. First, firefighters positioned three monitors to push flames from the dike fire away from the tank. One at a time, four super jet monitors were moved in to project foam directly into the tank. By 3 p.m. the fire noticeably diminished, but the heat of the tank shell prevented the foam from sealing against it.

By 6 p.m. the dike fire was out and only minor fire was left in the tank. The last of that died out at 11:30 p.m., about 60½-hours after ignition. The LASTFIRE report estimated the total loss at $22 million.

When Browning returned to his refinery, he recommended a reassessment of the facility's capacity to deal with a Milford Haven-size emergency.

"I had many discussions with the chief fire officer of the municipal fire department and our refinery management team who agreed with me that we needed to get larger foam monitors, improve the water supply and double the foam stocks available to us," Browning said. "Like Milford Haven, we were in an isolated location."

Today, the Fawley refinery has two Williams F&HC Six Gun portable nozzles, two miles of six-inch hose, two hose trailers and a 6,000 gpm portable diesel fire pump. Water to feed that equipment comes from large diameter water mains both above and below ground. Tapping into those mains are eight-inch manifolds complete with a six-inch valve/six-inch Stortz connection and eight 2½-inch fire service connections for the benefit of municipal responders.

"Today, we have the equipment on the market to deal with a jumbo storage tank fire," Browning said. "We now recognize the amounts of water we require and the foam concentrate we need."

Lessons Learned

  • No automatic fire detection and no fixed fire protection means no first line of defense.
  • Metal fatigue is not often a factor in storage tank failures, but the possibility should be considered in regions where high winds are common.
  • Boilovers give no warning. The only accurate way to track the downward progress of the heat is thermal imaging, a technology long accepted by municipal fire departments.
  • Milford Haven is a remote, isolated refinery. Resources traveled from far to reach the troubled facility. This distance must be factored into the emergency pre-plan for the refinery.

Editor's note: This after-action review appeared in "Disasters Man-Made" by David White and Anton Riecher that was published in 2011.