Catching Fire

David Pacchioli
January 01, 2000
fire burning yellow cellophane

Today is a burn day. The test cell, a ten-by-ten-foot square with corrugated steel walls and ceiling and a steel-and-concrete floor, feels chilly, even though its heavily reinforced door is open to the summer air. Just outside, a massive vine-covered bunker looms almost within reach, its 15-foot height blocking from view all but a swatch of blue sky. Butterflies flit on the vines. Inside, Matt Mench points into the middle of a dense thicket of metal pipes and foil-clad wires at a thick-walled stainless steel chamber like a small suitcase with two Plexiglas windows. "Take a look," he says. Behind the windows, which sit one atop the other, is a thin tube of aluminum, one-eighth-inch in diameter and 12 inches long, like a narrow drinking straw.

Mench, a Ph.D. student in mechanical engineering, and James Sturges, his undergraduate assistant, have been setting up for two days, getting everything perfect. In a few more moments, they will open the valves that send a stream of liquid oxygen, chilled to -140 degrees C, coursing into the system and up that skinny tube. Then they'll retreat to the next room, where Mench, after running down his list of last-minute checks, will push a lighted red button on a control panel. A bolt of current from two car batteries will ignite the frigid metal, and after two-tenths of a second in that pressurized oxygen-rich environment the aluminum will disappear, consumed like a candlewick by a flame that can reach in excess of 4,100 degrees C.

A super-high-speed camera will record this intense flash of fire. The camera, which cost $100,000 and shoots up to 11,000 frames per second, sits on a sturdy tripod, placed well back from the chamber and off to one side: its view of the windows is an image in a mirror.

"We don't want anything to happen to it," Mench says, squinting through the eyepiece. The windows routinely crack from the heat of combustion, he says. One test("our most violent case") burned through the chamber's two-inch stainless steel casing, and there are dark scars on the surrounding metal to prove it. "That was a $5,000 data point," Mench jokes.

The research he's doing here at Penn State's High Pressure Combustion Lab, a low-slung collection of cinderblock rooms tucked away behind the University flower gardens, could save a lot more than that. Indeed, the data already gathered, Mench says, should pave the way for vital safety advances in the air-separation industry.

Chemical companies, he explains, produce large quantities of gaseous oxygen, nitrogen, and other elements by distilling them from air. To do this, air-separation plants all over the world employ the same basic technology: pulling in air from the atmosphere, running it through filters to remove impurities, and cooling it to a liquid state by passing it through coiled heat exchangers. Then the liquid air is slowly reheated, so that each of its elements, having a different boiling point, reverts separately to gas, and can be easily collected.

These heat exchangers, Mench continues, are made of aluminum—because it's light-weight, readily available, and cheap. "It also has relatively good thermal properties," he says. At high temperatures, however, it is very reactive. "It burns well in oxygen, and releases a lot of energy."

Under normal conditions, this potential hasn't been a problem. Over the years, in fact, there have been very few safety incidents involving aluminum heat exchangers. "Back in the '60s, there was an explosion at an air-separation plant in Belle, West Virginia," Mench says, "but they thought it happened because they were pulling impurities in with the air—hydrocarbons from car exhaust, that sort of thing. So they changed their filtering process."

For the next 30 years, Mench says, there were no major problems. Then, three years ago, a massive explosion at an air separation plant in Fushun, China, killed "several" employees (the exact number is unclear). And on December 25, 1997, an earth-shaking blast leveled a similar plant in Bintulu, Indonesia. "Luckily," Mench says, "it was Christmas eve after midnight, and there were very few people around, so nobody was killed this time." In Bintulu as in Fushun, however, "the entire plant was destroyed in less than a second. What was left was mostly powder."

When Kenneth Kuo, now distinguished professor of mechanical engineering, arrived at University Park as an assistant professor back in 1972, he immediately set about finding a place for a high-pressure combustion lab. Kuo had earned his Ph.D. from Princeton analyzing rocket propellants, after working briefly in the aerospace industry.

plume of flame shooting from a metal cylinder
(Courtesy High Pressure Combustion Laboratory)

How propellants work: Hot gases squeeze through a nozzle. Being tested here is a propellant for deploying automobile airbags. In a collision, a sensor sends a volt to an igniter, which lights up the solid propellant. As it burns, the propellant produces gases, which inflate the bag.

At the time, he remembers, the only space available was a small two-room structure out on Big Hollow Road, behind the flower gardens. It was equipped with a test cell that had once been used by faculty in the chemistry department, but in recent years the building had been a storage shed.

"I had received funding from the Army to conduct some tests, and so I asked the College of Science whether I could use it," Kuo says. Permission granted, he began a research program that took off so quickly the University had trouble keeping a roof over its head. "In those days," Kuo says, "we even did experiments in the parking lot. We fired rocket motors horizontally into the earth bank."

In 1975, a much-needed expansion project tripled the lab's available floor space, providing room for three more steel-and-concrete-reinforced test cells. Funding continued from the Army, and more also came from the Air Force and the Navy. A second expansion took place in 1984, when Kuo arranged for the transfer of a 120-foot shock tunnel, the largest in any American university, from Princeton. "It had been run by my former professor there, Martin Summerfield," Kuo explains. "When he retired, it was broken up into pieces and put into storage. I requested that it be brought here."

A third expansion, funded by the Navy and the Chung-Shaw Institute of Sciences and Technology of Taiwan in 1988, added two large air-storage tanks, a compression room, and a supersonic wind tunnel for testing ramjets and ducted rockets. Finally, two years ago, through Pennsylvania's Ben Franklin Partnership program, in collaboration with Erie-based Hydro-Pac, Inc., Kuo acquired a high-pressure compression system that can deliver up to 30,000 pounds per square inch, enough pressure to simulate the inside of a gun barrel during firing.

Today, the lab is a cornerstone of Penn State's Propulsion Engineering Research Center, which in 1993 was established as one of two NASA Centers of Excellence for the study of ignition and combustion of "energetic" materials.

The core of what Kuo and his students do is basic research into the burning properties of new materials constantly being developed to power rockets, satellites, and military guns. Eric Boyer, a Ph.D. student in the lab, offers a short tutorial. "The inside of a rocket engine," he begins, "is a pressurized container of propellant." Propellant consists of a fuel—a substance that will release a lot of energy as it burns—and an oxidizer, without which the fuel can't burn in a sealed environment. Once the propellant is ignited, the oxidizer reacts with the fuel, and the hot gases generated by the combustion are squeezed out a nozzle, converting it into kinetic energy—the kind that blasts a rocket into space.

Burning rates differ widely—with the material being burned, of course, but also with pressure, temperature, and other variables. Liquid fuels (hydrogen is the most common) tend to be highly energetic; but they are also highly reactive, which makes storage and handling a problem. Solid propellants, on the other hand, are relatively inert: they can't react until ignited. In a typical solid propellant, fuel and oxidizer particles are mixed together in a substance that resembles tire rubber. Solids are easier to store for long periods, therefore, but they don't pack as much punch either. Hybrid designs currently being perfected combine liquid oxidizers with solid fuels in an attempt to achieve the best of both worlds.

Kuo's lab team actively researches all three types of propellants. Currently, Boyer and two other grad students, Yi-Ping Chang and John Mordosky, are testing a new class of liquid propellants based on hydroxy-ammonium nitrate, or HAN, a slightly modified form of the ammonium nitrate that is commonly used for fertilizer. HAN-based propellants are less toxic than traditional rocket propellants, Boyer explains, and unlike liquid hydrogen and liquid oxygen, they don't have to be maintained at very low temperatures. "They have lots of potential."

Nitromethane, another of the substances they are looking at, "might make a good replacement for hydrazine, the stuff that's used in thruster motors to keep satellites from drifting out of orbit," Boyer says. Unlike some HAN-based fuels, nitromethane has been around a while. "It's used in top-fuel dragsters, and model airplanes, and as an industrial solvent." And unlike hydrazine, which is a nasty carcinogen, nitromethane is essentially non-toxic. "It's a little acidic, is all," Boyer reports. "It has good properties, and it's cheap. It is sensitive to shock, however, so it has to be handled carefully. We're trying to understand it a little better. If we're going to use it in satellites, we need to know how it will behave over the long-term."

Boyer, Chang, and Mordosky conduct test burns in windowed stainless steel chambers called strand burners, similar to the one Matt Mench is using to burn aluminum. "The liquid is pushed up into the chamber by a piston," Boyer says, and then into a narrow tube made of quartz or another non-reactive material inside the chamber. "You want to feed the liquid in as fast as it's being burned." Tiny thermocouples—hair-like strands of chromium nickel and aluminum nickel—provide exact temperature measurements at different locations in and around the flame. Precisely spaced breakwires attached to the fuel tube determine the burn rate. "We time how long the flame takes to reach each one," Boyer explains. In some experiments, they also use an absorption spectrometer, which can separate out the wave-lengths of light a flame contains, revealing its chemical makeup. The entire burn—split-seconds in duration—is captured with high-speed cameras.

microscopic fuzzy beads
(Courtesy High Pressure Combustion Laboratory)

This scanning electron microscope image shows the bead-like burn residue from a solid propellant. The beads consist of mostly sodium chloride—salt.

Boyer pops a cassette into a VCR to show some highlights. First to appear are a succession of nitromethane burns, conduc ted at different feeding rates and pressures. The flame, in each sequence, is a narrow yellow band across the top of the clear tube; it burns smoothly and evenly like the wick of a turned-down oil lamp. At higher pressures, it simply glows a little brighter.

Another HAN-based propellant, known as XM46, provides more dramatic footage. "This one was developed by the Army for its liquid propellant guns," says Boyer. "It was not well-behaved." The first clip shows exactly what he means: Almost as soon as it's lit, the flame flushes violently downward instead of up, disappearing into the feeder tube and leaving a puffball of dark smoke spinning in its wake. Boyer smiles grimly. "We tested this stuff at different feeding rates, different pressures, using different reservoirs, different nozzle tips, we tried everything," he says. "For a year. At first we thought the feeding system, pumping from below, might be introducing agitation, so we tried a gravity system, but that didn't work either."

"Sometimes," Yi-Ping Chang reports, "the burn rate went down with increased pressure," the opposite of the expected. "It's not clear what's going on," Boyer says. What is clear is that XM46 has been shelved as a potential propellant until he and his colleagues can find out more. Currently, Chang is looking at computer simulations of the stuff. "I'm coming at it from a theoretical point of view," he says.

One of the lab's finest hours came in 1993, in response to the disastrous failure of an Air Force Titan IV rocket shortly after launch from Vandenburg Air Force Base in California. The Titan IV is the military's main vehicle for carrying communications satellites into space. In this case, scarcely a minute past lift-off, a Titan's solid rocket motor explo-ded, sending well over a billion dollars worth of rocket and payload crashing to the bottom of the Pacific Ocean.

The most likely cause of failure, according to an Air Force investigator, appeared to be a thin crack in the motor's solid propellant block, "probably made during a faulty repair procedure." Kuo, well-known for his expertise in the propagation of flame through cracks in solid propellants, was called in to consult. After convincing an Air Force panel that rapid pressurization in the motor's chamber could indeed cause a flame to penetrate a crack and cause an explosion, he returned to Penn State and ran tests under simulated Titan IV conditions, proving the theory. The efforts of Kuo and his students, according to an Air Force citation, resulted in a "record-setting recovery of launching status of the Titan IV rocket motors," and savings of hundreds of millions of dollars.

Rocket combustion continues to be the backbone of Kuo's research program. With the dropoff in defense funding during the late 1990s, however, has come a broadening of the lab's mission, to include studies of for various corporate clients. Matt Mench's work on the burning characteristics of aluminum, sponsored by the Allentown-based Air-Products and Chemicals, Inc., is one example. Another is a project by graduate students Grant Risha and Abdullah Ulas for Talley Defense Systems, an Arizona-based maker of combustion-related components, testing solid propellants for automobile airbags.

"There are three basic airbag technologies," Risha explains. The most common, and most reliable, contains a tiny pyrotechnic igniter, like a miniature spark plug, and a small quantity of a solid propellant. In a collision, an impact sensor sends a volt to the igniter, which lights up the solid propellant. As it burns, the propellant produces hot gases, which inflate the bag. "All this happens in 50 to 60 milliseconds," Risha notes.

Most airbags, he adds, use sodium azide-based propellants. "These put a lot of nitrogen in the exhaust, which is okay, but the manufacturing process is toxic and corrosive, and so is the waste residue." Another type of propellant, known as TAL 1308, might be cleaner, Talley's engineers thought. But they needed to know exactly how it burned. So Risha and Ulas burned it, in a strand burner setup simulating airbag conditions. They determined its temperature sensitivity and burn rate through a range of relevant pressures. They analyzed its gaseous products. On high-speed video they saw that as the propellant burned, a residue of tiny beads was cast from its surface. Scanning electron micrographs and x-ray diffraction showed these beads to be sodium chloride—salt.

Their conclusion, for now, is that TAL 1308 "burns well, and it generates okay gases. But we're not really here to determine whether this particular propellant is good or bad for use in an airbag. WeÆre just providing data on its behavior."

Back at the aluminum-liquid oxygen test cell, Mench and Sturges have started "pre-chill," flushing liquid oxygen through their test apparatus to cool it down before they get on with the morning's burn. "If we don't do this first, the liquid oxygen will boil off as soon as it hits those metal pipes," Mench explains. "That's how cold it is." In minutes, the pipes running into and out from the test chamber are white with frost.

schematic of rupture disk, windowed test chamber, aluminum sample
Courtesy High Pressure Combustion Laboratory

Part of a schematic diagram of the aluminum-liquid oxygen test cell (and control room) at the High Pressure Combustion Laboratory. In a windowed stainless steel chamber, in the center of the cell, a tube of aluminum, ignited by a bolt of current, burns like a candle wick.

Next Mench walks to a table in the adjoining room and grabs an old-fashioned standing microphone. His voice carries over the lab's indoor/outdoor P.A., announcing the impending test. ("Please stay clear of the fenced-in area.") Returning, he picks up a clipboard and scans down his final checklist. "Don't go past there," he says at last, indicating a line on the floor several feet from the door. Then he and Sturges take up their positions at two control panels facing the cell's front wall.

At Mench's signal, Sturges starts the camera; Mench pushes the button marked ignition. A slight pause; then a soft crack, a flash of light in the doorway, a whiff of something. "Smells like burned glass," Sturges says. "Usually," says Mench, "you don't hear that big of a pop."

A few minutes later, in a darkened conference room, Mench has threaded a projector with the film from a previous burn, and doused the lights. "Two hundred feet of film run through that camera in 1.76 seconds," he says. The strip he's set up, trimmed to show only the crucial instant, looks about a tenth that long. He runs the film in slow motion, narrating.

"This is test 109, one of our best examples," he starts. On the screen, dimly at first, appear the top and bottom windows of the test chamber. Then a soft light opens in the top window: the aluminum igniting. "With normal burning," Mench is saying, "you have relatively low luminosity;" and for a few slowmotion seconds, the tiny flame persists, tracking gradually down the thin shaft of metal. Then a tiny spark pops from the top, tumbles, lands near the bottom of the shaft. Another second passes, and phoom!—there's a roiling fireball, and the screen goes almost white.

Mench stops the film; plays it back. "You see what happens here? A piece of molten metal falls, this little speck of white. It lands, and ignites the outside of the aluminum, here. There's a pause, and then it burns through to the liquid oxygen. . . . " He shakes his head. "Looks like a volcano."

Violent Energy Release is what they call this phenomenon; Mench is the first researcher to capture it on film. "Imagine that happening to a threeton heat exchanger," he says. That's what obliterated the industrial plants at Fushun and Bintulu.

It took a while for Mench to come across violent energy release in the lab. For a year, another graduate student, Dave Johnson, burned the aluminum in an environment of gaseous oxygen, "the traditional way of testing metal combustion," says Mench. "With gaseous oxygen, you get a relatively slow, regular flame. The aluminum burns at a predictable rate."

The only trouble is, to get gaseous oxygen, aluminum heat exchangers have to chill air to liquid first then slowly reheat it. So, with the help of Air-Products engineers, Mench redesigned the test chamber. The makeover cost on the order of $100,000. What Mench found, however, made it worth doing. "In the liquid environment," he quickly observed, there's the potential for this sudden transition. When it happens, in less than a tenth of a millisecond, the burn rate jumps from half a meter of aluminum tube per second up to 40 meters per second, with a huge spike in the accompanying pressure trace. "High pressure means high heat release," Mench says, "so that spike is showing a sudden change in the rate of energy release."

It's showing an explosion, in other words. "This is not just a rapid boiling phenomenon, like what happens when liquid alu minum drops into liquid oxygen and causes rapid boiling and a pressure wave. Here's what we think happens. You have molten aluminum dripping, coming into contact with liquid oxygen. This causes violent boiling, which shoots a jet of oxygen directly into the flame zone." Once that happens, it's all over. "The thing is burning so fast it's self-propagating.

"This violent energy release can't happen in a gaseous-oxygen environment," Mench adds, "because the rate of diffusion—like the time it takes for a perfume to get across the room—limits the rate of burning. With a liquid, however,you've got this pool of high-density oxidizer to splash up. Fuel and oxidizer can be brought together very quickly."

Once they had proved the potential for violent energy release given a liquid environment, Mench and Sturges altered pressures and flow rates and liquid/gas ratios to see if they could establish a reliable set of boundary conditions for safe operating. After a year and a half, Mench reports at least preliminary success. "We've shown that violent energy release is predictable, to some degree. We've shown that you can separate regions of violent energy release and regions of no burning. What we need to do now is construct a refined map of that safety zone, one that will allow us to make recommendations." Perhaps such a map will convince the air-separation industry that violent energy release is real, and also avoidable.

"There's an awful lot invested in aluminum," in terms of industrial infrastructure, he explains. "Our sponsor, Air-Products, took some criticism from competitors early on for even suggesting there might be a problem. Even now, people will argue, 'The service record has been fine. What's the big deal?' Right now this is a very hot topic."

Kenneth K. Kuo, Ph.D., is distinguished professor of mechanical engineering and director of the High Pressure Combustion Laboratory in the College of Engineering, 140 Research Bldg. East, University Park, PA 16802; 814-863-6270; Matthew Mench, Eric Boyer, Yi-Ping Chang, Grant Risha, and Abdullah Ulas are doctoral students; John Mordosky is a master's student ; and James Sturges is an undergraduate student , all in mechanical engineering. The aluminum burning project is funded by Air Products and Chemicals, Inc., of Trexlertown, PA, under that corporation's Master Research Agreement with Penn State. For more information, visit

Last Updated January 01, 2000