Random Walk
The aftermath of the fire that engulfed a Boeing 737 when leaking fuel was ignited by a hot engine. Fortunately, no one was hurt.
Having a Blast
In 2007, soon after landing at Naha Airport on the island of Okinawa in Japan, a Boeing 737 started spewing fuel from a puncture in its right-wing tank. As the fuel flowed onto the hot engine, it ignited, causing several explosions as flames engulfed the plane. Fortunately, everyone evacuated in time and no one was hurt.
Of course, planes, trains, and automobiles don’t generally burst into huge fireballs whenever they’re near a heat source, whether it be a spark or a lit cigarette—despite what Hollywood might lead you to believe. But clearly, exploding fuel is a hazard, and understanding how it ignites allows engineers to develop the proper safety regulations to minimize danger.
Postdoc Sally Bane (PhD ’10), working with Joe Shepherd (PhD ’81), Johnson Professor of Aeronautics and professor of mechanical engineering, is analyzing how a spark can ignite flammable gas. She’s discovered that these spark-ignited explosions are a lot more complicated than previously thought, and her results bring much-needed updates to safety standards that are decades old—and, in some cases, based on data that are flat-out wrong.
For decades, engineers have determined the likelihood of a spark-ignited explosion in an aircraft by using a number called the minimum ignition energy (MIE). Each type of fuel or vapor has its own MIE value, and if a spark’s energy is below that value, then nothing should ignite. The problem, however, is that the MIE is based on data from old experiments, some from more than 60 years ago. Even though the numbers have been updated over the years, no one has ever gone back and reevaluated the original experiments in depth. The original data from the 1940s are still cited today, Bane says.
As an expert in explosions, Shepherd was recruited in 1996 by the National Transportation Safety Board to help figure out why TWA Flight 800 crashed into the Atlantic just 12 minutes after takeoff from New York’s JFK Airport (see “Learning from a Tragedy,” E&S No. 2, 1998). During the investigation, some of Shepherd’s experiments showed unforeseen variability in spark-ignited explosions—even with a spark below the MIE, the fuel sometimes still blew up. However, he didn’t get a chance to explore the issue further until Bane came to his lab as a graduate student in 2005.
According to Bane, performing this kind of experiment is difficult and time-consuming, and since there hadn’t been any obvious reason to doubt the old MIE data, people were content with the existing information. It took her five years to design, build, and run her experiments, which are among the most rigorous ever done.
Her explosions range from harmless puffs to tiny fireballs, safely contained in a solid-steel vessel weighing 200 kilograms. Inside the vessel are two pointy tungsten electrodes, separated by a couple of millimeters. Charge builds up in the electrodes, generating a voltage difference that ionizes the gas between them. The ionized gas creates a path for electrons to travel from one electrode to the other, and in just a few nanoseconds, you get a spark. The spark energies that Bane works with are relatively low—as low as 50 microjoules. You would need a million of these sparks to light a 50-watt lightbulb for a second.
A high-speed camera, which can take up to several hundred thousand frames per second, records the spark and any resulting explosion. Bane repeated the experiment for different spark energies and with various mixtures of flammable gases as suggested by the Aerospace Recommended Practices, standards developed by the Society of Automobile Engineers that guide the design and production of aircraft parts. The mixtures contained 5, 6, or 7 percent hydrogen, oxygen, and argon.
Unlike tests that are common in the industry, Bane’s are done in a sealed vessel, which means that she knows the gas’s composition, temperature, and pressure with a high degree of certainty. Furthermore, Bane says, in other setups, there’s no camera that directly observes the ignition. Instead, the only way to tell that there’s been an explosion is to watch whether a piece of foil, which is across an opening in the vessel, pops up—or, if the blast is violent enough, bursts open.
Bane discovered that there isn’t a single MIE value for a given flammable mixture; there’s no single energy that sets the threshold for whether an explosion is possible. Instead, she found, whether or not a certain spark ignites the gas is an exercise in probabilities, rising and falling with the energy of the spark. To be sure, at some point, the energy is too low and an explosion is impossible. But it’s not as cut and dried as people had thought.
In the case of the 7 percent hydrogen mixture, Bane’s results were roughly comparable to the experiments done by Bernard Lewis and Guenther von Elbe in the 1940s, which gave an MIE of 100 microjoules. She found that at 100 microjoules, the gas had a 10 percent chance of igniting. But for the 5 percent hydrogen mixture, the MIE given was 200 microjoules, while Bane determined that the spark had to be at least 780 microjoules before it even had a 10 percent chance of blowing up. It turned out that Lewis and von Elbe hadn’t actually had data for a 5 percent hydrogen mixture back in the 1940s—they had just extrapolated the MIE from data for the 7 percent gas.
This variability is likely an inherent characteristic of spark ignition, Bane says. Even when she kept the conditions as constant as possible, the explosions were never consistent. The spark itself—the channel of plasma connecting the two electrodes—is intrinsically irregular, wavering in shape and motion. Other complexities have been revealed in fluid-dynamic simulations that Bane and graduate student Jack Ziegler have run of the ignition process.
Bane and Shepherd are now working to develop better safety standards with Boeing, which funded the research. Last summer, she expanded her experiments to include kerosene, the type of fuel used in most commercial aircraft. Unlike clean-burning hydrogen, kerosene is a dirty fuel, so after each trial, Bane dons a protective jacket, Kevlar gloves, and goggles before opening the hot vessel to swab out the soot and set up the next experiment.
Bane is finding that it doesn’t take much energy to set off kerosene. She was able to ignite kerosene at 60 degrees Celsius with only 0.65 microjoules, while in Shepherd’s previous work, it took 40 microjoules to ignite the fuel at 52 degrees. “I anticipate that in future tests I’ll be able to ignite mixtures with even lower energies,” she says. The ignition of fuel vapor is highly sensitive to temperature change, Bane explains, and her experiment used shorter-duration sparks—about a couple hundred nanoseconds, which is a more accurate simulation of an electrostatic discharge—so the discrepancy isn’t a complete surprise. But the fact that it’s apparently much easier to ignite kerosene than the test mixtures specified by the FAA has certainly gotten the attention of the folks at Boeing.
Bane says her data can be applied to any situation where there’s a tank of fuel and the possibility of a spark—for example, in power plants, in the natural-gas tank that heats your house, or when storing hydrogen gas, which has garnered interest as a clean-burning fuel. But as for smoldering cigarettes blowing cars to kingdom come? “That could never happen,” Bane says. Gasoline needs to be extremely hot to ignite, and a cigarette just won’t cut it. But it sure looks cool. —MW
Below: A spark between the two pointy electrodes ignites the 7 percent hydrogen mixture, generating a fireball. It’s not as dangerous as it looks—the explosion is small. For comparison, the distance between the electrodes is only a few millimeters. The snapshots are taken at 5, 50, 100, 175, and 250 microseconds.
