Author(s): Adam St John. Published on September 1, 2017.

The Science of Flame Jetting

How can a container of flammable liquid suddenly jet flame like a blowtorch when it encounters an ignition source? Why do so many people, many of them children, continue to be injured in these incidents? And how can a simple, inexpensive device protect against this hazard?


In September of 2015, NFPA Journal published a cover story entitled “Hey Kids, Watch This,” on the potential dangers of scientific experiments involving flammable liquids and open flames in classroom laboratories and other educational settings. The story, written by Andrew Minister, a chief fire protection engineer at Pacific Northwest National Laboratory in Richland, Washington, and chair of the technical committee for NFPA 45, Fire Protection for Laboratories Using Chemicals, detailed a number of incidents where students were injured, many seriously, in accidents that occurred during science demonstrations.

Related Content

Flame Jetting 101
Learn what variables can affect whether a flame jet will occur and how long it will be.

'A long rope-like finger of flame'
Read an NFPA staffer's harrowing account of witnessing a flame-jetting incident.

Anatomy of a Flame Jet
As images from ATF testing demonstrate, a lot can happen in a third of a second.

“Over the last 15 years, according to media accounts, scores of students have been injured or burned in dozens of these demonstrations; the actual number of incidents, as well as the number of students injured, are likely much higher,” Minister wrote. “Many of the injured students suffer second- and third-degree burns on their faces and upper bodies. The burn injures are very painful, the recovery is long and agonizing over many months or years, and victims can be scarred physically and mentally for life.”

A month after that story appeared, in October, three students were severely burned during a fire incident in a high school science lab in Virginia.

The teacher was demonstrating the “rainbow experiment,” an exercise that uses a flammable liquid to showcase the effect of various chemical elements on flame color. As the fire died down, the teacher rekindled the flame by pouring out additional liquid from the container. As she did so, a large flame was expelled from the container, engulfing five students and severely burning three of them. The flame event also ignited a backpack and a chair, resulting in fire sprinkler activation in the classroom.

The lab where I work, the Fire Research Laboratory (FRL) at the Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF), had previously conducted testing on the phenomenon known as flame jetting, and Virginia investigators were aware of our work. They immediately suspected flame jetting was to blame for what had occurred in the science lab, but they needed more information to be certain. Fairfax County fire investigators contacted our lab and requested additional testing to evaluate their fire origin and cause hypothesis.

The ATF FRL conducted 49 tests using ethyl alcohol in various types and sizes of containers, including glass and plastic containers in one-liter and one-gallon sizes. (The container in the Virginia chemistry lab demonstration had a one-gallon capacity.) All containers had “necked” openings, and none included nozzles. Flame jetting occurred in all 49 of those tests. Investigators regularly witnessed flame jets in excess of 15 feet during the FRL testing, results that were consistent with witness statements and the documented fire damage in the classroom. Virginia investigators described the scientific data as “invaluable” and the testing validated their ignition hypothesis in the fire origin and cause report. No criminal charges were filed against the science teacher. (My thanks to Kerwin McNamara, battalion chief with the Fairfax County, Virginia, Fire and Rescue, for sharing case information used in this article.)

Flame jetting isn’t limited to laboratory or educational settings—the phenomenon has been observed in a variety of situations and settings, in some cases with fatal results. A solution to this hazard exists in the form of flame arrestors, simple devices that are essentially thin metal or plastic screens that typically cost less than 50 cents. Flame arrestors have been shown in testing to prevent flame jetting from portable flammable liquid containers that would otherwise produce jets in certain conditions.

The Consumer Product Safety Commission is currently working with researchers (including ATF) and the American Society of Testing and Materials (ASTM) to develop the first standard to evaluate the effectiveness and durability of flame mitigation devices for portable flammable liquid containers. The purpose of the standard is to ensure that flame mitigation devices meet minimum effectiveness criteria and remain intact after repeated use prior to widespread installation in these types of containers. The new ASTM flame mitigation device standard is expected to be completed sometime this year.

The need for research

Our initial research into flame jetting began seven years ago when Michigan investigators charged a man with the burning death of his six-year-old daughter. The father was allegedly pouring diesel fuel—a detail that was later determined to be inaccurate—onto a fire in a fire pit when his daughter became engulfed in flames, but he couldn’t explain exactly how she was ignited. “She was at a safe distance. I was pouring the container and all of a sudden, there’s just fire everywhere. I don’t know what I did,” the man told investigators. Law enforcement officials decided that more information was needed, and asked the ATF Fire Research Lab in Maryland to perform forensic fire testing and collect scientific data related to the case. The ATF FRL is the world’s largest forensic fire research facility and is tasked with conducting experiments in support of fire investigation and fire research. Many of these experiments require constructing a portion of the fire scene in the ATF FRL’s burn rooms and conducting a series of fire tests.

The father speculated the flames from the fire pit may have entered the container, which did not have a nozzle installed, creating a momentary flame jet that engulfed the child as she passed by the fire pit. There was no documented thermal damage to the standard two-gallon portable flammable liquid container (made by Blitz® Manufacturing, model #50810), which was found on the scene with approximately a tenth of a gallon of liquid remaining in it. A sample of the liquid found in the container was analyzed and determined to be not diesel, but rather a mixture of gasoline and heavy petroleum distillate. The father stated that the liquid in the container was collected as he purged contaminated fuel pump lines as part of his job servicing fuel pumps. He estimated that the container had less than a half-gallon of fuel in it at the time of the incident.

With no relevant public research available, 17 fire tests were conducted at the FRL’s Large Burn Room. The tests included flammable liquids—those with flash points of less than 100 degrees F, such as gasoline—and combustible liquids, with flash points at or above 100 degrees F, such as diesel fuel. For safety purposes, an apparatus was constructed that allowed a portable flammable liquid container to be remotely poured from an enclosure. None of the tests included containers with nozzles. The container was positioned above a natural gas tube burner that served as the pilot flame. The first four tests consisted of 100 percent gasoline, while the remaining 13 tests were conducted with a mixture of gasoline and diesel, or all diesel.

In the first gasoline test, the portable flammable liquid container was tilted over the flame, and without warning a large flame jet rapidly developed from the mouth of the container and extended in excess of 2 meters (6.5 feet). The container was not damaged. Overall, flame jetting was documented in 13 of the 17 tests. Flame jetting was observed in tests using 100 percent gasoline and a gasoline/diesel mix, but it was not observed in any of the tests using 100 percent diesel. (The ambient temperature must be above the flashpoint of the liquid to support combustion and flame jetting; the all-diesel tests never jetted, as testing occurred below diesel’s flash point.) The longest jet observed extended over 4.5 meters (15 feet) horizontally, resulting in sustained head-to-toe burning of a mannequin placed at the same distance the child had been from the fire pit in the Michigan incident.

Flame jetting was only observed when the container was being tilted and the vapors were pouring from the mouth of the container. It is hypothesized that initially, when the container is upright, the head space above the liquid is too fuel-rich and above the upper flammability limit, meaning that combustion is not supported within the container. As the container is tilted and vapors begin to pour from the open mouth, however, air is entrained into the head space and the fuel-rich mixture eventually falls within the flammable limits; if an ignition source is present and combustion occurs, the flame propagation condition inside the container can lead to flame jetting.

Flame jetting was also documented with burner flames in direct contact with the mouth of the container (piloted ignition) as well as with burner flames that were several inches from the mouth of the container at the time of ignition (non-piloted ignition). Testing indicated that flame jetting typically occurred less than five seconds after liquid began to pour from the mouth of the container. The entire jetting event lasted less than one second, with no observable warning signs prior to the phenomenon. When jetting did occur, there was no evidence of thermal or pressure damage to the container. The length of the flame jet was dependent on several variables, including the total quantity of liquid, the mixture ratio, and the percentage that the gasoline was “weathered,” or evaporated.

On this point, flame jetting was only observed with weathered gasoline—25 percent weathered gasoline by weight was evaluated. No flame jetting was documented when “fresh” gasoline was poured from the container. Fresh gasoline releases vapors more readily than weathered gasoline at the same temperature. Consequently, it is hypothesized that even when air is entrained into the container headspace while pouring, fresh gasoline releases flammable vapors readily enough that the headspace never drops below the upper flammable limit and therefore does not support flame propagation within the container. Weathered gasoline, by comparison, is slower to release vapors and can support flame propagation inside the container, leading to flame jetting.

The ATF FRL testing results supported the argument that the daughter’s death in the Michigan case was likely a tragic accident. The father’s inability to explain the flame-jetting event was consistent with the rapid development and unpredictable nature of the flame-jetting phenomenon. The prosecutor dropped the murder charges and the father pled guilty to misdemeanor child endangerment charges. The self-evident nature of the ATF FRL forensic tests prevented a likely innocent person from being charged for a crime he did not commit.

As a result of this case and others, as well as continued incidents in educational settings—including one in May at a Texas pre-school, where a dozen three-year-olds suffered flame-jetting burns—the hazard of flame jetting is receiving wider attention. Earlier this year, a bill was introduced in Congress designed to reduce the number of people injured in these incidents. The Portable Fuel Container Safety Act of 2017, currently in committee, would require flame arrestors on portable flammable liquid containers.

Flame arrestors have been shown to be highly effective at preventing flame jetting. Liquid from the portable flammable liquid container can be poured through the arrestor’s screen, but if an ignition source is present and combustion occurs, the device prevents the flame front from traveling through the mouth of the container and into the headspace above the liquid. The ATF FRL conducted seven tests with a flame mitigation screen in the mouth of the same portable flammable liquid containers used in the previous flame-jetting tests. The flame-arrestor screens prevented flame jetting in all seven tests. Flames never propagated beyond the location of the screen.

Flame mitigation devices were successfully utilized for decades in mining lamps, and more recently they’ve been used in mufflers, high-proof liquor bottles, and in some commercial flammable liquid containers. Despite being highly effective and relatively inexpensive, the vast majority of consumer-grade portable flammable liquid containers are not currently equipped with the device, which the bill in Congress aims to reverse.

ADAM ST. JOHN, P.E. is a licensed fire protection engineer at the Bureau of Alcohol, Tobacco, Firearms and Explosives’ Fire Research Laboratory in Ammendale, Maryland. Top Photograph: Thinkstock