As fracking turned the United States into a major producer of natural gas over the past decade, federal regulators approved the construction of export terminals along the Atlantic and Gulf coasts while relying on industry safety calculations that critics say significantly understate the potential force of a specific type of accidental explosion.

The particular event that worries engineers outside the business has a very low probability of happening but could have exceedingly destructive consequences if it does.

Under new leadership since January, the federal Pipeline and Hazardous Materials Safety Administration, or PHMSA, told The Washington Post it intends to draw up rules at some point next year that would deal with the risk in question.

The move comes as two new plants just outside Brownsville, Tex., are poised to begin construction. Yet each has already passed through the safety permitting process.

The danger is one the United States more or less backed into. It developed with the rush to build export terminals with equipment to liquefy natural gas, and an early assessment that these facilities were no more hazardous than the terminals built to import the stuff back when the United States consumed more gas than it produced.

That was a mistake.

Eventually, regulators and industry engineers came around to the understanding that these terminals do pose inherent new dangers, almost as an afterthought. But even to this day, federal regulators accept at face value the industry’s calculations regarding what engineers call a vapor cloud explosion.

Critics here and abroad — engineers — have argued since 2016 that those calculations seriously underestimate the destructive potential of such an event. Over the years, those arguments have been enhanced by greater specificity and greater documentation, even as the Trump administration pushed hard for more U.S. export terminals and more sales of American LNG, or liquified natural gas.

The danger is not what you might suppose: It’s not the natural gas, though that also poses risks. The threat of a vapor cloud explosion comes from the heavier hydrocarbons an export terminal relies on to chill the natural gas so deeply that it turns into a liquid, which is then loaded onto ships for sale abroad. These hydrocarbons are called refrigerants, and under the right conditions, a major leak of volatile refrigerants on a windless day could lead to the buildup of a cloud of ground-hugging vapor, until a spark sets it off.

The model used by the industry, and implicitly accepted by PHMSA, has held in case after case that the destructive force of a vapor cloud explosion would be spent by the time it reaches the perimeter of whatever terminal is under consideration. But specialists not associated with the gas business say this is wrong.

One study by British experts found that a hypothetical vapor cloud explosion could be up to 15 to 20 times as powerful as what the planners modeled.

They have examples to back up their argument, mostly from facilities that handled petroleum products similar to those used in natural gas liquefaction plants. Vapor cloud explosions are rare, but at least three have occurred in the United States and Latin America since 2009.

That year, one in Puerto Rico demolished a tank farm; the remains looked like Stalingrad. The blast registered as the equivalent of a small earthquake. There were no fatalities, but 300 nearby homes were damaged. The vapor in that case was from gasoline, which is less dangerous than the refrigerants used for natural gas.

An explosion in Venezuela in 2012 destroyed an oil refinery and inflicted heavy damage on a nearby army barracks and several residential areas, killing 47.

Vapor clouds have exploded in Algeria and Britain. One in Philadelphia led to the permanent closure in 2019 of what had been the largest oil refinery on the East Coast. The explosion catapulted a 38,000-pound vessel across the Schuylkill River.

“If something doesn’t get corrected, there might be some terrible accidents,” said Jerry Havens, the recently retired director of the Chemical Hazards Research Center at the University of Arkansas, who has been sounding the alarm ever more urgently over the past five years. “I can’t get anybody to do anything about it.”

Developers calculate explosion risks

Jared Hockema is the city manager of Port Isabel, Tex., a community of 6,000 that relies on sport fishing and commercial shrimping for its livelihood. From his office just steps from Laguna Madre, the bay that separates the mainland from South Padre Island, it is a mile and a half back through town to the Brownsville-Port Isabel Highway.

There, just behind the H-E-B Supermarket, on flat reclaimed scrubland between a lake named Vadia Ancha and the Brownsville Ship Channel, a company called Texas LNG wants to build a big new export terminal.

Port Isabel has been fighting every step of the project, before state and federal regulatory bodies and in the courts. The city is worried about air pollution and fires. It is concerned about interference with shrimpers’ livelihoods, and certain that a big plant will drive away tourists. Hockema said the site is a habitat for ocelots, an endangered species. And then there is the risk of explosion, overshadowing what is for the most part a community of color.

“A facility like this threatens our way of life,” Hockema said. Because Port Isabel is mostly poor, and mostly Brown, he said, “they think they can get away with more here than in other parts of the country.”

“This is a blatant example of environmental injustice,” said Rebekah Hinojosa, a Sierra Club activist who works with a group called Save RGV, referring to the Rio Grande Valley.

PHMSA has approved the safety plans for three proposed LNG export terminals along the Brownsville Ship Channel, though other regulatory hurdles remain. One was canceled this year when its developer, Annova, decided that the faltering export market wouldn’t justify the investment. Hockema, Hinojosa and their allies hope the remaining two, Texas LNG and Rio Grande LNG, which would abut each other on the Brownsville Ship Channel, meet the same fate before they can be built. But not so far.

During the permit process, all three projects presented calculations regarding vapor cloud explosions based on small potential spills and the model that critics say lowballs the shock wave emanating from a blast. Hockema points to the danger of debris from Elon Musk’s SpaceX launch site, just across the channel, falling on the terminal sites, potentially puncturing pipes or tanks. (When a prototype rocket crashed in March, Musk tweeted that the cause was a small methane leak. SpaceX did not respond to a request for comment.)

Texas LNG has the endorsement of Sen. Ted Cruz (R-Tex.). Its founder and chief operating officer, Langtry Meyer, accompanied President Donald Trump on a mission to China in November 2017 to try to drum up business.

A spokeswoman for the company, Camilla Siazon, said Texas LNG would have no comment and referred questions about explosion hazards to federal regulators.

Rio Grande LNG is a subsidiary of a Houston-based energy company called NextDecade. “Rio Grande LNG hired third-party experts to perform modeling using highly complex, globally recognized hazard modeling software tools,” Patrick Hughes, a senior vice president, wrote in an email. “The modeling was performed consistent with federal requirements and provides reliable determinations of the potential hazards used for facility siting.”

Employees of PHMSA, the pipeline safety agency, knew in at least 2016 that the issue of vapor cloud explosions needed to be considered in greater depth, thanks to Havens’s warnings. For most other hazards, federal standards for calculating risk are spelled out, but regulators had not taken that step with vapor cloud explosions — and still haven’t. They accept what the project developers provide.

Deflagrations vs. detonations

As the 2016 election approached, the agency, jointly with the British Health and Safety Executive, sponsored an engineering workshop on the topic. A series of slides in that workshop demonstrated the hazards.

A typical export terminal might have 50 tons of refrigerants on site, consisting of some combination of ethylene, propane, isobutane, isopentane or hexane. A leak at a moment when there is no wind is the most dangerous because the vapor that forms as the liquid evaporates won’t disperse. It will gather in a cloud that grows until the leak stops or all the liquid spills. The participants reviewed vapor cloud explosions at different sorts of facilities and made the point that various heavier-than-air hydrocarbons, including gasoline, act in similar fashions and can be used for modeling risks.

One difference, though, is that refrigerants are more volatile than gasoline and exist naturally in a gaseous state, so up to 100 percent of a leak could be expected to form a vapor cloud.

The explosion in San Juan, Puerto Rico, occurred 26 minutes after the gasoline leak there began and involved 78 tons of vapor. The leak in Amuay, Venezuela, was slow but kept leaking for more than an hour before the vapor detonated.

The key finding was that the force of the explosion in such cases is not greatest at the center, dissipating outward, but generally homogeneous within the cloud, especially if it is contained by outer barriers, as is typically the case.

The model used by the industry does not account for that. Plans are drafted and calculations made that suggest a maximum “overpressure” — or shock wave — of one pound per square inch would be felt at the peripheries of these terminals. That’s enough to break windows. But the 2016 workshop suggested that overpressures can be far greater than that — with buildings destroyed and widespread fatalities.

PHMSA officials shelved the workshop report, and nothing came of it. Drue Pearce, who was named acting administrator when Trump took office a few months later and then served as deputy until this past January, said she had never heard of the workshop or the report.

Tristan Brown, newly appointed as acting head of PHMSA after working as an aide to Sen. Amy Klobuchar (D-Minn.), defended the dedication of the agency’s staff and said new rules on vapor cloud explosions are scheduled to be drawn up in the next year, as part of nine safety-related rules on the agency’s pipeline regulation agenda.

In a formal response provided to The Post by the agency, PHMSA officials said: “The agency keeps safety at the forefront as we review LNG facility applications and operations — including factoring in the most up to date hazard modeling — and ensuring operators are doing the same. Our work over the last few years on VCE [vapor cloud explosions] has continued and evolved with broad input from researchers and stakeholders.”

The statement continued: “Regulations are one tool in PHMSA’s toolbelt to ensure safety in the design and operations processes — and under new leadership we have been immediately focused on building out our capacity to more quickly develop and implement new regulations. This is particularly important given the dozens of new regulatory mandates passed in the bipartisan PIPES Act of 2020 (last December) — with updates to the LNG facilities rules being one of our top regulatory priorities.”

The dynamics of a vapor cloud as it forms and spreads, and the calculation of overpressures, can seem dauntingly arcane. Engineers even have different words for differentpotential explosions, which are determined by the amount of vapor involved, and act differently. Industry calculations are based on weaker “deflagrations.” Critics worry about much stronger “detonations.”

Vapor clouds with a diameter exceeding 100 meters, or about 110 yards, are especially dangerous, based on a survey of a representative selection of incidents, HSE, the British agency, said in a statement provided to The Post. It is difficult to find examples of such a large cloud that did not explode. “There may be an element of underreporting, especially where no one is killed, but it still seems appropriate to work on the basis that there is a fairly high probability of explosion for these very large vapor clouds,” it said.

The statement suggested that the model used to assess explosion risks is not fundamentally flawed but is only as good as the assumptions made by its users, regarding not only the size of the vapor cloud but its turbulence, the speed with which it forms and the probability of ignition, among other factors.

“Experience has shown,” the statement said, following on a 2019 study, that different users “can produce quite different results for the same scenario.”

The model is a subset of a software program called FLACS, for Flame Acceleration Simulator, developed by a company called Gexcon. The sub-model for vapor cloud explosions has been dubbed Q9, and this is what has drawn critics’ attention.

Vincent Tam of Warwick University in Britain was the lead author of a paper critical of Q9 published this year in Eng, a peer-reviewed journal.

“The results in this paper showed that Q9 systematically underpredicts” the force of vapor explosions, he wrote.

In a follow-up email, Tam wrote that industry risk assessments usually do not consider windless conditions because modeling them is too difficult. (Both the San Juan and Venezuelan explosions occurred in what engineers call nil-wind situations.)

PHMSA said it is reviewing Tam’s paper.

Gexcon points to studies by former company engineers that support the Q9 model. One of those engineers, Filippo Gavelli, used Q9 in writing the vapor cloud sections of environmental impact statements for several proposed LNG terminals. He now works for Blue Engineering and Consulting, in Ellicott City, Md.

“Every hazard modeling tool that Blue utilizes has been widely used worldwide and accepted by numerous regulatory agencies,” Gavelli wrote in an email. “Blue’s work products comply with all federal regulations and applicable codes and standards.”

Through his company, Gavelli has a $473,000 PHMSA contract to develop an assessment tool for evaluating vapor cloud models — Q9 among them. The contract, he wrote, was awarded with an eye toward revising federal regulations to require specific hazard modeling.

PHMSA, in its statement, said Blue won the contract because it was the sole bidder.

LNG terminals are typically designed with vapor barriers at the perimeters. These are intended to keep clouds of vapor from drifting off site and potentially toward an inhabited area.

“This is a double-edged sword,” Tam wrote. “The barrier stops flammable gas migrating into surrounding buildings leading to multiple explosions.”

But when the wind isn’t blowing, he wrote, a flammable cloud “can accumulate and eventually fill up the entire facility enclosed by these barriers.” This occurred in an accident in Jaipur, India, in 2009. “There was a leak of gasoline, its vapor was trapped by boundary walls. The resultant vapor cloud explosion completely demolished the entire facility.”

Havens, the recently retired director of the Chemical Hazards Research Center, is more blunt about the installation of vapor fences.

“It’s really stupid,” he said. “It’s a really bad mistake.”

A 1990 Energy Department study, gathering dust for decades, came to the same conclusion.

Slumping demand

Since the 2016 workshop that spelled out the dangers of a vapor cloud explosion, six export terminals, with facilities to chill and liquefy natural gas, have begun operations, in Texas, Louisiana, Georgia and Maryland.

Pearce, the former PHMSA deputy administrator, wrote in an email to The Post that there was never a hint of pressure from the Trump White House to expedite the permitting of LNG export terminals. (The final permits are issued by the Federal Energy Regulatory Commission, but PHMSA signs off on the safety features.) “My first priority was and is safety; I don’t believe in taking shortcuts,” she wrote.

None of those six terminals has experienced a vapor cloud explosion.

Other proposed terminals, including the two at Brownsville, are in the permitting process. But customers are becoming difficult to find amid a global slump in demand, and some projects have been canceled, including the Annova terminal in Brownsville and one proposed for Jordan Cove, Ore. Last month a Canadian export terminal was canceled, and Sempra Energy said it would push back construction of a new terminal in Port Arthur, Tex., at least a year.

In Port Isabel, Hockema said he fears that the Texas LNG terminal will be approved and built, and then abandoned for lack of business, leaving the town with a permanent, hulking eyesore. That would be unfortunate for his city, but not as unfortunate as the detonation of dozens of tons of volatile hydrocarbons.