TWA 800: another anniversary—Part 2
Masterpiece of misinformation
I imagine I am there, I am the plane. I am it; it is me. I am cradling the fuselage holding 230 passengers and crew, and the four gigantic Pratt and Whitney engines, the wings, the nose, the tail, the redundant hydraulic systems, the aeronautical marvel that was the world’s first jumbo jet, the story of whose unlikely ascent was related with consummate authority and grace by Clive Irving in “Wide-Body: The Triumph of the 747,” (William Morrow, 1993).
I will lift it up to where it is safely out of the way of whatever was fired from below that destroyed it and left 230 bereaved families, which was seen by 12-year-old FBI interviewee Adam Coletti, a passenger in U.S. Air Flight 217, flying at 21,000 feet as TWA Flight 800 crossed beneath it. Adam said he saw the wake of a boat and the shape of a boat, which seemed to speed up. Then he saw a redness, “like it was blinking,” he said to me at his home in Cranston, RI. “It was 10 to 15 seconds after I saw the red that I saw the explosion,” which “went up from the boat--just really quick,” he said. We don’t know whether the FBI took Adam’s account seriously enough to follow up and inquire about the presence of a large vessel visible to an airplane passenger at 21,000 feet, and what it was doing there, nine miles south of the South Shore of Long Island, out where recreational boats rarely go.
The NTSB, which for its part concluded that TWA 800 was destroyed because of an explosion of fuel vapor and air in its center fuel tank probably caused by a short circuit outside of the tank, has never revealed the identify of several nearby surface vessels caught on radar, one of which was steaming at 30 knots (about 35 mph) close to the flight path of TWA 800. To explain the fact that this vessel did not change course, or slow down, or make an emergency radio call to the Coast Guard when the plane exploded above and behind it, the NTSB theorized that nobody on board would have been aware of the explosion or fireball or falling debris. That seems highly unlikely. Was it the same boat Adam Coletti apparently saw? Chairman Jim Hall of the NTSB was evidently thinking of that vessel when he asked an NTSB staffer something about it during the five-day hearing into its investigation that the NTSB held in Baltimore in December 1997. The FBI’s Jim Kallstrom, who told a packed press conference a few weeks earlier that his agents had combed marinas in search of any boat that might have provided a platform for a terrorist attack, never mentioned the vessel moving at 30 knots that had been picked up on aircraft radar. I filed a Freedom of Information Act request for any information identifying the vessel, and the NTSB sent me a “no records” response, which I appealed, but I don’t believe I ever received anything more regarding that request.
The TWA Flight 800 disaster was unique: in a record of aviation fuel tank mishaps compiled by the FAA, and one compiled by the NTSB, no previous airplane in flight, powered solely by Jet A aviation kerosene fuel as TWA 800 was, had been brought down because an electrical fault caused a fuel/air explosion in its center fuel tank; and none has happened since.
Joe Sutter, who led the team that conceived and developed the 747, said by phone in 1999 from his home in Kauai, “When we designed the airplane we did every damn thing we could to make a fuel tank explosion not happen…I was the chief engineer on the 747, and you go to bed nights thinking about what you’ve done that day to make sure it was the safest thing you could do when you’re designing that part of the airplane—.”
Is it possible, though, that a four-year, multi-million dollar investigation missed a significant clue?
Or was the safety board being selective about which bits of evidence it probed?
Remembering the explosion that destroyed the Paris-bound Boeing 747 on the evening of July 17, 1996, as she had heard it on Long Island, a witness described the sound to me as “a thunderclap.”
“It was very loud, like a shock wave,” recalled another witness, at a gathering of eyewitnesses at a hotel on Long Island held to mark the fifth anniversary of the crash. He further recalled a “firework,” a white smoke plume and a fireball before he heard the explosion. Two more witnesses told the FBI they heard “a loud boom,” which one likened to “an M-80 firecracker.” The NTSB’s final report notes that many witnesses heard a loud sound that they described as being like a boom.
A retired government expert estimated that such a loud sound, heard at that distance, would be equivalent to one ton of a high explosive like TNT going off.
The NTSB reported that the 747’s cockpit voice recording ended with a brief, very loud sound.
All of this seems to suggest that the explosion was very loud indeed.
But when NTSB investigators compared a visual printout of the very brief sound recorded by TWA 800’s cockpit voice recorder, called the “sound signature,” with those recorded by the CVRs in other air disasters and incidents, such as PanAm 103 in 1988 (plastic explosive) and a United Airlines incident (mechanical failure), they judged that the TWA 800 sound signature was closer to the sound of the mechanical failure than it was to PanAm 103 or to an Air India flight that had also been brought down by a terrorist bomb. Yet at 117 milliseconds duration, the TWA 800 sound appeared much briefer that the sound of a mechanical failure (a cargo door separated in flight).
And several sources suggested that a fuel tank explosion within the belly of a big jet would not be audible nine miles away.
As part of its investigation, in the summer of 1997 the NTSB set off an explosion inside the center fuel tank of a retired Air France 747 at a former Royal Air Force airfield in England, at Bruntingthorpe in Leicestershire; but when I visited Bruntingthorpe while working with a team on a proposed BBC TV documentary in early 2002, someone who had been no more than a mile from the site described the sound of the explosion as a “dull thud.”
Michael G. Zabetakis, a prolific U.S. government researcher into the flammability of gases and vapors, told me in an interview not long before he died that he wouldn’t expect to hear a fuel tank explosion if he was more than a few miles away.
Zabetakis’s Jan 24, 2005 obituary in the Pittsburgh Post-Gazette was headlined “Michael G. Zabetakis: Renowned Explosives Expert.” Wikipedia claims that one of his reports was for decades one of the most widely cited sources of flammability data for more than 200 gases and vapors.
In that report, “Flammability Characteristics of Combustible Gases and Vapors” (U.S. Bureau of Mines, 1965) Zabetakis explained that what we call explosions can either be deflagrations, which are episodes of usually fast burning, most often associated with flammable mixtures of gases or fuel vapors; or detonations, which are usually much more rapid and violent reactions associated with high explosives like TNT or Semtex. The technical distinction between the two reactions, Zabetakis said, is that in a deflagration the initial reaction moves more slowly than the speed of sound, whereas in a detonation the process moves faster than the speed of sound.
The NTSB reported that its Bruntingthorpe center fuel tank explosion was, as described in a separate report by British government observers at the event, “a fairly slow deflagration.” That seems hardly comparable with the less-than-two-tenths-of-a-second duration of the very loud sound that marked the end of Flight 800.
And at Bruntingthorpe the NTSB used, not Jet A, the widely used jet fuel that airlines, including TWA, voluntarily adopted back in the 1960s because it was deemed safer than an earlier more explosive fuel linked to several fatal civilian airline accidents and many military plane accidents, but propane, a highly explosive flammable gas that belongs to the same family of hydrocarbons as Jet A but which has caused many camping and domestic accidents when not handled carefully. It is considered so dangerous that supermarkets bar customers from bringing empty canisters into the store.
Yet even with a propane/air mixture, the NTSB was not prepared to risk using an electric spark. Ignition was achieved by “detonating an explosive device”— inside a tank in which the NTSB had sealed shut the passageways leading to vents at the wing tips, presumably to prevent any of the propane escaping before someone was ready to flip a switch to set off the explosive device to ignite the mixture.
Why propane and not Jet A? A note in the NTSB report mentions propane was chosen in part because of the cold climate; but it was late July in England—midsummer, when temperatures are often in the high 70s or above.
While it is considered technically possible though highly unlikely for a deflagration in a mixture of fuel vapor and air to transition to a detonation, Zabetakis observed that the ignition energy necessary to initiate a detonation in such a mixture “is usually many orders of magnitude greater than that required to initiate a deflagration.”
Detonation also requires a significantly more concentrated mixture of fuel vapor and air than a deflagration.
But the immediate question facing NTSB investigators was not whether the mixture of fuel vapor and air inside TWA 800’s center fuel tank might have detonated; it was whether the mixture was flammable at all.
Flammability 101
Liquid fuel must be heated and evaporate to form a vapor in a volume of air before it can be ignited and burn or explode. As in the jet engines themselves, it is the vapor/air mixture, rather than the liquid fuel itself, that ignites and burns.
Zabetakis reported that, after ignition, a flame must continue to burn, independently of the ignition source, or the mixture cannot be called flammable. A spark might produce a momentary ignition of a vapor/air mixture, after which, if the temperature is too low and there isn’t enough fuel vapor present to sustain the combustion, the flame could simply sputter out. This phenomenon was first identified in a 1952 study in which Zabetakis participated, which discussed how an ignition source could raise the temperature and pressure in its immediate vicinity and produce a brief flame in a mixture that contained too little fuel vapor to be called truly flammable (“Research on the Flammability Characteristics of Aircraft Fuels,” G.W. Jones, M.G. Zabetakis, et al, Wright Air Development Center, 1952).
Navy researcher L.J. Nestor counted a fuel/air mixture as flammable only if the flame traveled the entire four feet of a transparent vertical Pyrex tube, which was four inches across, with an igniter at one end. If the flame died after two feet or even three, Nestor would record the mixture as non-flammable (“Investigation of Turbine Fuel Flammability within Aircraft Fuel Tanks,” Naval Air Propulsion Test Center, Philadelphia, 1967). “A condition was classified as flammable only if the flame traversed the complete four-foot length of the combustion chamber,” he wrote.
Nestor cited two previous studies that used a similar method to establish flammability limits: Zabetakis’s 1965 work and an older study from 1952 (“Limits of Flammability of Gases, and Vapors,” H.F. Coward, G.W. Jones, U.S. Bureau of Mines).
Nestor writes, “The principle of visually defining flames is widely accepted and has been used quite extensively for determining limits of flammability.” He did a validation test on his four-foot tube apparatus by using it to investigate the flammability limits for a simple gas, comparing the results with the limits established by previous studies done on the same gas. He found “good agreement.”
Nestor calls his method “the classical approach for defining flammability limits of gases…,” and he credits Zabetakis: “Zabetakis defined a flammable fuel/air ratio as one where the flame propagation is essentially independent of the ignition source.”
But the NTSB briefly considered Nestor’s tube method thus:
“The most commonly used method of all, the flammability limit tube developed at the Bureau of Mines, (Coward and Jones 1952; Zabetakis 1965; Zabetakis et al. 1951), has never been standardized,” stated a May 1999 NTSB report into spark ignition energy measurements in Jet A. Evidently Nestor’s method did not meet the NTSB’s needs.
That 1999 report, published by the NTSB as Exhibit 20T, states that of several factors to be considered relative to TWA 800, “fuel temperature is the most important since the fuel vapor is created by evaporation of a liquid fuel.” Yet the report does not mention the temperature of the liquid fuel again.
And in what seems almost like a mockery of the known circumstances that prevailed in the 747’s center fuel tank, 20T’s spark ignition tests were conducted in a vessel with a very small (1.84 liters, or about one half gallon) volume inside which the temperature of the atmosphere was precisely controlled by electrically powered heating pads attached to the outside walls of the vessel.
There’s no hint that the NTSB heated the fuel up to what it considered a reasonable temperature, given the conditions at JFK on 7/17/96 and the warming effect of the air conditioning machines beneath the tank, which investigators blamed for heating the tank, and waited to see if enough of the fuel would evaporate to make a flammable mixture in the huge space above the fuel, which in any fuel tank is called the ullage. In fact, in 20T the vessel was preheated before the fuel sample was even introduced. To further help keep up the temperature, the whole thing was enclosed in a wooden box! The safety board’s valiant attempts to control the temperature inside its half-gallon vessel would effectively prevent the vapor from condensing again and returning to its liquid form, something that is known to be a feature of fuel tanks. Pity the poor, massive, maligned center fuel tank, having to draw heat up, no doubt inefficiently, from the air conditioning machines on its own in order to play its assigned role, without the ingenuity of the NTSB to help keep that temperature up—and in such a cavernous space too.
The report says that these tests and another set of tests in a larger volume vessel (1180 liters, or about 312 gallons) showed that it was possible to ignite Jet A vapor, producing a “propagating flame.” But the report does not define “propagating.” The report says that at least 6,000 volts was needed “to cause spontaneous electrical breakdown across the gap.” The maximum in any of TWA’s wiring near the center fuel tank was considerably less than 1,000 volts. The report discusses sparks with a duration of less than 5 microseconds—that’s 5 millionths of a second. By contrast, Nestor reports using a spark of about one half second duration. That’s 100,000 times longer than a 5 microsecond spark. If Nestor needed a half-second spark to thoroughly test his mixtures at the limit of flammability, what might he have said had someone in his lab suggested exposing those mixtures to a spark whose life was much briefer than the blink of an eye?
In one perhaps inadvertently revealing statement, 20T admits that the NTSB was well aware that the vapor/air mixtures it was testing as representative of the conditions inside TWA 800’s center fuel tank at the altitude of the explosion were either at the very limit of flammability, or else beyond that limit.
The statement concerns “vapor pressure,” which varies with temperature and among flammable liquids. For example, the vapor pressure of JP-4, the more explosive fuel I mentioned that Jet A replaced in commercial aviation 60 years ago, is much higher at much lower temperatures than the vapor pressure of Jet A. The higher the vapor pressure, the more explosive the fuel.
The vapor pressure of a fuel “will in turn (Shepherd et al, 1997) depend primarily on the fuel temperature,” 20T says. But in referring to a graph, it also notes that “The vapor pressure used in this comparison was obtained at a high mass loading (400 kg/m3) and is representative of what might be obtained in a flashpoint/type test, but not for a nearly empty tank such as in TWA 800.”
The NTSB presented measurements in systems unfamiliar to many Americans, e.g. temperatures in Centigrade not Fahrenheit, and weights and measures in metric units—kilograms not pounds, cubic meters not gallons. The NTSB used “mass loading” to mean the weight or mass of liquid fuel within a volume of air in the tank—I think. According to the NTSB, 400 kg/m3, or 400 kilograms per cubic meter, meant 400 kilograms of liquid fuel in a cubic meter of the tank. The NTSB characterized 400 kg/m3 as a half-full tank, and 200 kg/m3 as a quarter-full tank.
The 20T ignition tests were done with the equivalent of 200 kg/m3 of Jet A (quarter-full tank) and 3 kg/m3 of Jet A (nearly empty tank—the TWA 800 situation), inside the half-gallon and 312-gallon vessels. In other words Shepherd and his colleagues were choosing to try to ignite mixtures whose strength and concentration were, by their own admission, far below what would typically occur in a test designed to determine the flashpoint, which is the lowest temperature at which the vapor above a sample of liquid fuel can be briefly ignited.
The NTSB suggests that the 3 kg/m3 ratio might result in decreased Jet A vapor pressure, “leaner fuel/air mixtures in the ullage and possibly higher ignition energies.”
But 20T has already conceded that the vapor pressure at 200 kg/m3, let alone 3 kg/m3, would be significantly below the predicted vapor pressure in a flashpoint test, suggesting that these mixtures were unlikely to be flammable.
So when 20T reports that Caltech’s Joe Shepherd and his colleagues were able to ignite those mixtures, we have to wonder: Would the mixtures be able to pass Nestor’s four-foot tube test, or would they be susceptible to the phenomenon noted in the Jones and Zabetakis 1952 report, of an ignition source briefly raising the temperature and pressure in its immediate vicinity?
Also, regarding the ignition energy needed at the limit of flammability, a 2019 report into the lower flammability limits of Jet A and some alternative aviation fuels, by a team at Purdue University in Indiana, noted “It is well known that near the flammability limits, the minimum ignition energy required to ignite a mixture is extremely high”—too high, perhaps, for the NTSB to be able to plausibly suggest that the 747’s own systems, no matter how badly they malfunctioned, could have supplied a sufficiently strong ignition source.
The NTSB claimed that “at 14,000 ft [roughly the altitude of the explosion] the mixture should be flammable for any temperature higher than about 30 degrees (C).” That’s 86 degrees (F). This would be about 30 degrees below the flashpoint of the fuel in TWA’s center fuel tank at JFK. By way of comparison, a record kept by TWA’s fuel contractor on the afternoon of July 17, 1996 pegs the temperature of the fuel loaded into the 747’s wing tanks at 78 degrees (F). Jerry Biscardi of Ogden, the fuel contractor, confirmed to me in a phone call that indeed the fuel Ogden had delivered to the TWA aircraft whose next flight was to be Flight 800 was at that temperature. The flashpoint drops with altitude but not as low as 86 degrees (F). The notion that at only eight degrees warmer than the fuel Ogden delivered that day the fuel would be capable of generating a flammable vapor that if ignited would threaten the safety of TWA 800 and everyone aboard, seems preposterous. The NTSB did not go into the relationship between the fuel temperature and the temperature of the vapor/air mixture above the liquid fuel. Nestor reported that a Jet A vapor and air mixture at 87 degrees (F), when exposed to a high energy spark, produced so little reaction that his instruments were hardly able to measure it. Nestor noted a phenomenon: the lower flammability limit dropped significantly when erratic manouvers or turbulence shook up the liquid fuel, producing a spray. These conditions did not occur during TWA 800’s 12 minutes or so of flight.
Nestor points out that a precise analysis of the atmospheric conditions inside any aircraft fuel tank during flight presents insuperable practical difficulties; yet this is essentially what the NTSB was asking the victims’ families, media and general public to trust it to do regarding the condition of TWA 800’s center fuel tank at the altitude of the explosion.
The NTSB conducted another series of ignition tests using Jet A vapor inside a “quarter-scale” tank, carefully heated, whose walls were pre-strengthened to withstand an explosion. It seems possible that this may have been intended to save the NTSB embarrassment if the reaction was too feeble to do much damage to the walls. As it was, this series of tests produced lackluster results: many times the flame failed to travel to all the interconnected bays and often just died. The tank’s volume was actually only one 64th the volume of a full-sized tank, which the NTSB acknowledged in a footnote.
NTSB metallurgist Jim Wildey asserted at Baltimore that, despite the extreme rapidity of the event, the breakup sequence—the actual order in which the plane’s structure came apart—was apparent to those who studied the fracture surfaces, and that it led, first, from inside the center fuel tank and thence into the fuselage, which split into two pieces. But Wildey conceded in his account that cracks that he said occurred at a later stage of the breakup sequence were happening essentially simultaneously with things earlier in the sequence that supposedly led to them.
Shepherd in a Caltech alumni magazine article mentioned a 20 pounds-per-square-inch reaction inside the tank—which an NTSB staffer at Baltimore had agreed would have been sufficient to damage the tank structure, leading immediately to the plane’s disintegration. Shepherd figured out the surface area of the tank’s inside surfaces and, having done the arithmetic, said that 20 psi would produce an enormous force of half a million pounds—far more than the tank was built to withstand.
However, several people with some knowledge of physics told me that this was misleading; 20 pounds per square inch means that 20 pounds press on each square inch, and no more. But another factor is that the pressure spike measured by Shepherd was already sharply reduced after one second, yet one FAA engineer who contributed some calculations estimating that somehow a huge force of thousands of psi would have been focused within the center fuel tank upon a very strong structural beam beneath the tank that snapped, told me that the force necessary to do that damage would have needed to be applied for “way more than one second.”
The possibility of a vapor/air mixture igniting but producing a weak reaction was well known to researchers. “Aircraft fuel tanks…may safely contain low-order reactions, with and without venting.” (“Aircraft Mishap Fire Pattern Investigation,” Joseph M. Kuchta and Robert G. Clodfelter, Aero Propulsion Laboratory, Wright-Patterson Air Force Base, 1985.)
But the explosion that destroyed TWA Flight 800 was anything but a “low-order reaction,” as shown by the sound heard by witnesses on shore and also by some radar data released to me by the FBI under the Freedom of Information Act that shows that extremely fast-moving debris was ejected from the 747 in the first seconds after power was lost, suggesting that the explosion was violent and sudden.
Safety Board investigators would doubtless protest that each of its tests was carefully designed to explore a particular aspect of a center fuel tank explosion they already knew had happened.
Regarding that, in researching the documentary I mentioned, we learned of a metallurgist who had caused a headache at Calverton, the old airfield on Long Island where the recovered debris was taken and where part of the 747’s fuselage was reconstructed. Her name was Lenise Keskinel, and she was part of the TWA team that had a role in the investigation, because under NTSB rules TWA was a designated party. Keskinel apparently found some evidence on a recovered fragment of the tank that she did not believe could be easily explained as a consequence of an explosion inside the tank, and she would not shut up about it, and she was eventually sent home to TWA’s Kansas City base. Keskinel’s competence and professionalism impressed Capt. Jerry Rekart, who led TWA’s team investigating the crash, who told me he was disappointed when she was transferred back to Kansas City.
I attempted to find Keskinel, who had since left TWA, with no luck. I discovered that she was a cousin of actor Robert DeNiro’s then-wife, Grace Hightower, but I didn’t succeed in reaching her to ask her to help me contact her cousin.
Jerry Rekart met with BBC director Christopher Olgiati, former CBS News producer Ted Landreth and me at a diner off I84 outside Danbury, Conn., in 2001 to discuss the investigation and our proposed film. Rekart said that following an article I wrote, about a report by a young TWA pilot, ex-Navy, who had seen a ship he had recognized as a guided missile cruiser or destroyer off Gilgo State Park on the afternoon of July 17, 1996, he (Rekart) had asked an NTSB staffer whether a naval ship was indeed in those waters that day and had received the answer “Probably.”
Was the TWA 800 explosion then a deflagration or a detonation? At Baltimore and in the reports I have read, the NTSB did not say.
End